AMD quietly cuts FX-, A-series prices

You might not know it from looking at the date stamps on AMD’s processor pricing pages, because they’re still a few months old. A quick comparison of the existing pages versus those stored in Google’s cache reveals the truth, though: prices have dropped across nearly the entire FX and A series of processors.

Processor Old price New price Difference
FX-8150 Black Edition $245 $205 20%
FX-8120 Black Edition $185 $165 12%
FX-6200 Black Edition $165 $155 6%
FX-6100 Black Edition $145 $135 7%
A8-3870K Black Edition $135 $115 17%
A8-3850 $135 $105 29%
A8-3820 $129 $115 12%
A8-3800 $129 $105 23%
A6-3670K Black Edition $115 $100 15%
A6-3650 $115 $95 21%
A6-3600 $109 $95 15%
A6-3500 $85 $75 13%
A4-3400 $71 $59 20%
A4-3300 $66 $55 20%

The hardest hit were the FX-8150, A8-3850, A6-3650, A4-3400, and A4-3300, all of which got their prices slashed by 20% or more. The FX-4100 and FX-4170 were unaffected, and AMD looks to have taken the A6-3620 out of the picture altogether—it’s no longer listed in the new price list.

Based on our performance numbers to date, I’d say these cuts are overdue. Take the FX-8150, for instance, which fell slightly below Intel’s $220 Core i5-2500K on the performance scale despite being priced at $245. Or the A8-3850, which failed to keep up with AMD’s old Phenom II X4 840 overall despite costing a good $20-30 more. In some cases, the new prices seem to be what they should have been all along.

Of course, it seems clear why AMD has made these cuts: because the retail debut of Intel’s new Ivy Bridge chips is imminent. That’s unfortunate. If the numbers we got out of the Core i7-3770K are any indication, then other Ivy Bridge offerings might also be slightly quicker dollar-for-dollar than their predecessors. And if that’s the case, then AMD might be left charging slightly too much for some of its chips yet again. (Thanks to TR reader Jones for the heads up.)

Comments closed
    • clone
    • 11 years ago

    lol, and “bother” is inflammatory vs curious which was the motivation.

    • Chrispy_
    • 11 years ago

    [herp] Save yourself the guesswork and click on his profile [/derp]

    His watercooling setup is approaching money-no-object territory. I haven’t watercooled for ages, but the specs on his waterblock say it can dump 400 watts with ease. At only 1 gallon a minute, that block can dissipate 400W at under 15°C T-delta, so I would say he’s for real.

    • Vasilyfav
    • 11 years ago

    When the highest tier CPU is $205, they honestly don’t deserve the FX- moniker.

    FX- should basically be reserved for the “supercars” of the CPU world. Like the original FX series FX-74 or the like

    • Chrispy_
    • 11 years ago

    [url=http://games.adultswim.com/robot-unicorn-attack-twitchy-online-game.html<]Hmmm. I don't think I need Bulldozer for this :\[/url<]

    • Bensam123
    • 11 years ago

    Same for someone called NeelyCam.

    • Bensam123
    • 11 years ago

    Does it bother you that it bothers him?

    • NeelyCam
    • 11 years ago

    You’re in a bad mood an awful lot for someone with your screen name

    • NeelyCam
    • 11 years ago

    Does it bother you?

    • clone
    • 11 years ago

    why are you responding to yourself?

    • clone
    • 11 years ago

    I’d pay $100 for the top one…. even 2nd in line… wouldn’t hesitate for a moment.

    • OneArmedScissor
    • 11 years ago

    I think your link is all floating point. But that’s not to say that there’s a plethora of things that benefit from 8 integer cores.

    The bigger problem is that, whether they work or not, a big pile of cores and cache in lieu of less with lower latency is generally not a good idea fora PC.

    Bulldozer isn’t necessarily flawed or “faking” anything, but grossly inappropriate in the day and age of application-specific CPUs.

    • maxxcool
    • 11 years ago

    Oh look, fanboi cant take reality…

    • maxxcool
    • 11 years ago

    I was in a bad mood friday, damn fanbois’s looked like they needed whipping…

    • maxxcool
    • 11 years ago

    “” Each module has two integer cores that are wrapped in some shared resources, including the front-end instruction fetch and decode units, the FPU, and the L2 cache and its associated data pre-fetchers “”

    Its not just the fpu though. locking a l2 cache and prefetch really hampers the unit. And pretty much whacks any idea of each “int core” running at the same time. And if it does, just like TR’s review, and Anand’s review…. you end up with the other thread on the “hyper thread core”… and not a true full cpu module.

    unless they are going to add another fpu, or split it up in pile driver, this is not nor ever will be a 8 core cpu.

    • flip-mode
    • 11 years ago

    You’ll pay for it over the long term with the higher power consumption. And you’d still only be getting Core i3 2100 level of performance in the majority of applications. Bulldozer is a rotten lemon, so even the lemonade tastes nasty.

    • flip-mode
    • 11 years ago

    I’m betting he’s one of the 99.9% that are. If I’m wrong I’ll happily stand corrected.

    • ronch
    • 11 years ago

    Funny how, even after these price cuts, they’re still being sold at slightly higher prices (particularly the FX-8150). Price cuts, assuming there’s no threat from Ivy Bridge, would indicate lack of demand. Actual pricing though, suggests a shortage. So it looks like there’s a lack of demand and a shortage at the same time. Is that possible?

    What’s happening here is that AMD is doing these price cuts so the general trend among e-tailers is price drops as well, even if they don’t exactly match AMD’s SRPs. Obviously this is meant to spur demand. However, the fact that e-tailers are hiking prices up a bit would indicate a shortage of some sort. So AMD wants to spur demand when actual supply is still a bit shaky. Ok…

    EDIT – I think AMD has to do another round of price cuts real soon if they want their products to make sense. They’ve got to wake up to reality and stop relying on their fan boys to buy their stuff at stupid prices. They’ve also got to get the motherboard situation corrected, not just in the U.S., but everywhere. Where I live, I can’t get a good AM3+ board. It’s either too expensive (it’s like adding insult to injury amidst already overpriced FX chips), or too cheap and barren.

    • Bensam123
    • 11 years ago

    They use similar if not identical pricing for their OEM products. OEMs usually sell their products at MSRP to customers and get a volume discount for being a OEM. The pricing information is still relevant.

    • ermo
    • 11 years ago

    I am tempted to point out that being an ‘anal geek’ — as you so eloquently put it good sir — is what makes technology work in the first place.

    In math, stringent accuracy is at the very core of mathematical reasoning. In accounting, accuracy is the basis for checking and showing the health of a business and inaccuracy is considered a liability. In marketing, accuracy is less important than fooling your target audience into believing your hype and buying your product.

    In this case, a 20% price cut sounds better than a ~16.3% price cut and that’s the end of the story as far as I’m concerned.

    • ermo
    • 11 years ago

    I get what you are saying and I agree, but in the larger scheme of things, I wonder if the pricing for DIY (enthusiast) parts is as important as we make it out to be, given that us DIYers probably constitute a fairly small percentage of the overall amount of CPUs sold?

    If Llano and Brazos had been failures, things would probably be looking a lot more grim for AMD, though I do wonder what kind of profit AMD is turning on any given FX part after the recent price cuts. Assuming that FX CPUs are always sold at a profit, perhaps the price cuts are based on a projection along these lines:

    Assuming that p is profit (p > 0), p – c is profit after cuts (p – c > 0) and n is units sold where 0 < n[sub<]before[/sub<] / n[sub<]after[/sub<] < 1 (more units are sold after the cuts) n[sub<]after[/sub<] * (p - c) > n[sub<]before[/sub<] * p (more overall profit through larger volume) You also have to wonder what it's like for an AMD employee to read the comments we make about their business acumen and engineering skills :S ... EDIT: A preview button would be nice, TR 🙂

    • Anonymous Gerbll
    • 11 years ago

    If the table said % change, then yes, i would agree. However it is the correct equation for % difference which is cited in the table and the table is not referring to a change even if the article itself mentions change. The table would need to be labeled as such.

    • Anonymous Gerbll
    • 11 years ago

    That would be correct if the table was for % change. The wording of the table is ‘Difference’ however.

    • Bensam123
    • 11 years ago

    Check Newegg… that selection isn’t all that thin. You’re just riding this out based on assumptions of past trends regarding this specific area.

    You don’t NEED to overclock to make what I’m saying lucrative either, nor are registered dimms even required. It’s another area to look at that is generally being ignored for the same stigmas you’re presenting right now, regardless of how true they actually are.

    • kc77
    • 11 years ago

    I can say that if you walk into a BD build with your eyes wide open in a lot of instances BD is quite competitive. When you factor in pricing of server builds they are in orders of magnitude cheaper.

    It really isn’t that bad of a theory Bensam is making. As of right now I’m burning in my Dual C32 build (god was it a headache probably one of my toughest) but I can safely say that so far (I’m about an hour into the burn in) that it’s meeting my expectations and exceeding them.

    First to get a SB DP core around the 2500/2600K area you are looking at $800+. To get 8150 like performance (which does compete with the 2500/2600 you are looking at $350 which is what I paid for my 4284’s which I think is the sweet spot in the Opteron line.

    Second, as I said before if you buy BD for what it’s good at most won’t be disappointed. I do a lot of VM work and video encoding. So for me BD makes sense. The slower memory doesn’t matter much since i5/i7/8150 aren’t memory constrained that much in the DDR3 spec. In the old days a 33Mhz increase on SDRAM brought unbelievable performance. This isn’t as much of a problem now so the decrease in memory speed is hardly noticed for most tasks.

    To get a idea of the performance. My previous build was an i7 920 overclocked to 3.0GHz. I used the Matrix as a test. A 1080 transcode under handbrake on that build was about 4 hours… 5 if I was using the system. As of today this is now down to 1 hour. So 4x the performance of the old build. Not bad.

    • CuttinHobo
    • 11 years ago

    AMD’s version of Tick-Tock? They’re doing it wrong. 🙁

    • SPOOFE
    • 11 years ago

    Unless you happen to regularly engage in a work load that requires unicorn sparkles and/or cherries on top.

    • just brew it!
    • 11 years ago

    He didn’t specify that he was using air cooling…

    • flip-mode
    • 11 years ago

    Um, No. There is no air cooler capable of dealing with 400 watts – that would not even be possible with entry level water cooling; it would take a very high-end water cooler to soak up 400 watts from 400 mm^2. Air coolers can’t deal with much more than 200 watts. The Thermaltake Frio claims to be able to handle 220 watts, but that’s all extremely dependent upon multiple variables – like the T-delta required to be able to dissipate 220 wats – i.e., what must T-max and T-ambient be in order to dissipate 220 watts, and what must the contact area be.

    • just brew it!
    • 11 years ago

    Mother of God… how the f do you cool a 400W CPU? I assume you had a watercooling setup…?

    • Waco
    • 11 years ago

    Power constraints are really what hold them back. My 8120 when I had it was pulling around 400 watts FOR THE CPU ALONE when overclocked to 4.5 GHz and loaded down 100%. Motherboards that can reliably deliver that kind of power aren’t all that cheap, not to mention the amount of heat dump you get from running that kind of power through a CPU.

    Comparatively, my 2600K pulls nearly half that when cranked up to 5 GHz. 🙂

    • tfp
    • 11 years ago

    Actually from TR’s own bench marks the FX scales much better than intels chips with 8 threads however the chip is so weak from a single threaded stand point it has no chance of catching up.

    So I would say their version of hyperthreading is pretty good just AMDs chips aren’t “wide” enough plus other issues…

    • NeelyCam
    • 11 years ago

    You hang out with us because you love us. You’ve said so many times

    • JLW777
    • 11 years ago

    Alexko is correct, as an accountant I just can’t stand by as this numbers game gets more muddled up and maybe it will costs someone a job or make them look like a tit in front of their employer.

    I think I shouldbreak this down to different scenarios:

    1.) If the 8150’s old price is $205 and the management wants to increase it to $245, then (205*X=245 > X = 245/205 = roughly 20 %

    2.) If 8150’s old is is $245 and $205 (Which it is in this case) either u can do the approach of directly finding the discounted percentage (245-205)/245 * 100 = 16.3% (rounding difference) or find out the using basic algerbra the proportion to original prices (245* X = 205 > X = 205/245 = 0,8367 > 1 – 0,8367 = % differences

    1.) = Mark up
    2.) = Discount

    thats where the author mixed up. Assumed the cost of the chip is $205 but it is a new selling price.

    P.S I review junior’s work in analytical reviews and I pick up at least 2 a week. So it is a common misconception.

    • just brew it!
    • 11 years ago

    Server boards generally lack overclocking/tweaking options (a big strike against them for enthusiasts). Registered DIMMs reduce performance and drive up total system cost (another strike against server boards for both enthusiasts and consumers). While I agree that low-end server boards do exist, the selection is pretty thin; and given the feature set, they just don’t make much sense unless you’re actually building a server or high-reliability workstation.

    Unless you’re focusing on the tiny subset of enthusiasts and consumers who can make good use of 16 slow-ish cores (instead of a smaller number of faster ones), the idea is a non-starter.

    • JustAnEngineer
    • 11 years ago

    And there’s the answer. It’s pretty rare these days that I will upgrade a processor without also upgrading the motherboard.

    My Socket-FM1 system works very well in its HTPC role. When I’m ready to upgrade, I’ll buy a new processor and motherboard. The old Socket-FM1 package will be trickled-down to a family member and their old Socket-939 motherboard and CPU will be retired.

    • JustAnEngineer
    • 11 years ago

    An uneducated workforce that cannot do basic math is one of the things that holds back economic development.

    This therefore appears to be both a third-world and a first-world problem.

    • yogibbear
    • 11 years ago

    I’m not sure you’re either feeling sad or sarcastic, I can’t quite tell. 😛

    • lycium
    • 11 years ago

    only 8 cores for $200? i’ll sell you 26 BogoCores (TM) for just $40! throw in another $10 and i’ll even make the dials go to 11.

    *sigh* and i thought people had learnt from the ghz-race days not to measure performance by a single number… i guess that’s why everyone thinks gpus are magically faster at everything: they just have so many more “cores”!

    • Bensam123
    • 11 years ago

    Just because they’re made for servers or even stamped with the name ‘server’ does not necessarily imply they’re ungodly expensive. You can find them right around $200 for C34 or G34, duel sockets will add an extra $100.

    $200 is well within a price range for enthusiasts and a bit on the high end for consumers, but probable.

    • Goty
    • 11 years ago

    Anything perceived as pro-AMD is evil, hence the downvotes.

    • sweatshopking
    • 11 years ago

    WHY ARE YOU NERDS ALL SUCH ANAL GEEKS?! % CHANGE IN TEMP?! GRAPHS? COME ON!

    edit: here it goes again. downvoting for not being insane. idk why i hang with you nerds so much. you’re all sooo damn geeky, and sensitive about it!

    • ImSpartacus
    • 11 years ago

    Ok smartass, let’s take a closer look at how percentage change is defined.

    Let X and Y be real numbers such that X>Y.

    I know three ways of measuring percentage change.

    The first is the percentage decrease. We can call it a discount when X is the initial price of a good and the price lowers to Y. This is the percentage given in ads (i.e. 20% all steaks!). It is given by:

    [b<](X-Y)/X[/b<] =X/X-Y/X =1-Y/X or [b<]1-Y/X[/b<] =X/X-Y/X =(X-Y)/X Two lines of work (that a middle schooler could do) show that these two 'formulas' yield the same result. The next is the percentage increase. We can call it a mark-up when Y is an initial price of a good and the price increases to X. This is what you would use if an initial price increased (e.g. "Beans got marked up 30%!"). It is given by: [b<](X-Y)/Y[/b<] =X/Y-Y/Y =X/Y-1 or [b<]=X/Y-1[/b<] =X/Y-Y/Y =(X-Y)/Y Again, there are two identical forms. Note that percentage decrease [b<]is not[/b<] equal to percentage increase and we [i<]must[/i<] be told which type of percentage change is being used. Cyril used percentage increase in a situation where a normal person expects to see percentage decrease (and dubiously called it "Difference"). The final measure of percentage change is a red-headed step child. Nobody can even agree on a single name. I've seen people call it "percentage difference" and "average percentage change" and a bunch of other things. Only accountants and financial people use it because their numbers can go arbitrarily up or down and they don't want to run into the problem that Cyril suffered in this article. They use the average of percentage decrease and percentage increase to provide one measure that just measures the 'difference'. It is given by: [b<](X^2-Y^2)/(2XY)[/b<] Remember this just an arithmetic average of the first two forms. It's a cake walk to show it: [(X-Y)/X+(X-Y)/Y]/2 =[(X-Y)Y/(XY)+(X-Y)X/(XY)]/2 =[((X-Y)Y+(X-Y)X)/(XY)]/2 =[(X-Y)Y+(X-Y)X]/(2XY) =(X-Y)(X+Y)/(2XY) =(X^2-Y^2)/(2XY) And just to show you that there's a reason financial folk use this kind of percentage change, we'll find the absolute error between percentage increase and percentage decrease: error = first thing - second thing =(X-Y)/Y-(X-Y)/X =(X-Y)X/(XY)-(X-Y)Y/(XY) =[(X-Y)X-(X-Y)Y]/(XY) =[(X-Y)(X-Y)]/(XY) =[b<](X-Y)^2/(XY)[/b<] So all of Cyril's 'Difference' numbers are exactly too high by exactly that error amount. Just let X be the old price and Y be the new price. As you can probably guess, the greater the change, the greater the error between the two methods. Since AMD barely dropped their prices, the error is kinda small. But it's still tacky as hell.

    • just brew it!
    • 11 years ago

    Problem is, you still need a server motherboard since it is a different socket. So while the CPU may be in the consumer/enthusiast price range, the total platform cost is not. (Though it [i<]does[/i<] make it possible for you to build a totally kick-butt server on the cheap...)

    • flip-mode
    • 11 years ago

    It’s pretty bad if it gets spanked by a hyperthreaded dual core in the majority of things.

    • Goty
    • 11 years ago

    ITT: People who don’t really know what hyperthreading is.

    • flip-mode
    • 11 years ago

    Saving works in mysterious ways.

    • Goty
    • 11 years ago

    It’s more for the GPU than the CPU. The lower-end Llano models have trouble enabling all of AMD’s lovely shader-driven video processing features, so I’d want to go with the 6550D. Also, I do a bit of light gaming on my HTPC, so the stronger GPU couldn’t hurt.

    • just brew it!
    • 11 years ago

    Just like a non-hyperthreaded Athlon64 performed better than a hyperthreaded P4 back in the day. It’s all just marketing…

    • Chrispy_
    • 11 years ago

    Not really worth arguing about;

    Bulldozer could have 16 cores and [i<]unicorn sparkles with a cherry on top[/i<], but unless it can beat an intel chip for the same money, it's irrelevant.

    • Chrispy_
    • 11 years ago

    Huh? Where are you getting your figures from?

    [url<]https://techreport.com/articles.x/21813[/url<] In no game test is the 8105 faster than an i5-2400, and it's usually struggling to beat an i3-2100 Same for the particle simulations Same for the panorama photostitches Same for the PiColor tests Same for video encoding Same for map compiling Rendering and scientific computing are a wash - the 8150 wins some slightly, loses others more heavily to the 2500K, but the two chips are going to be approximately matched here. In fact, the only clear win for Bulldozer is 7-zip, and this victory is overshadowed by the fact it's a very AMD-biased test in which, I quote, "The FX-8150 is only slightly quicker than the lowly Athlon II X3 455" As if that wasn't damning enough, Bulldozer draws more than double the power of the 2500K, whilst largely losing to it in benchmarks, and of course, won't overclock anywhere near as easily as your typical 2500K. I don't know where you got the idea that the 2500K was "a good bit slower in multithreaded tests". This site and these articles pretty much debunk that theory. I will agree that in [b<]some[/b<] of the most heavily-threaded situations, Bulldozer beats the i5 by a non-trivial margin, but even now - a year after launch, none of the promised windows scheduling patches have really helped and in the overwhelming majority of cases, the vast majority of people looking for a new processor would get better everyday performance from one of the cheaper, lower-clocked i5's. In [b<]some[/b<] situations the Bulldozers are very fast. In [b<]many[/b<] situations the Bulldozers are [b<]DEPRESSINGLY SLOW[/b<]. Like '5-years-ago' slow where they're way down in the charts fighting against budget offerings with performance lower than Intel's original, first-ever true quad-core chip, the Kentsfield; January 2007.

    • ish718
    • 11 years ago

    You are partially wrong.
    There is 2 INT schedulers per module, so it can perform 8 INT operations at once.
    There is only one FP scheduler per module, so it can perform only 4 FP operations at a time.

    • Deanjo
    • 11 years ago

    I realize that, just pointing out there are reasons to go with an 8150 over an 8120.

    • NeelyCam
    • 11 years ago

    High margins are essential for a company that consistently hemorrhages money due to one-time costs.

    But AMD doesn’t have high margins because its products have low performance/cost. BD is big, fabbed on a badly yielding process, and it’s performance is low compared to SB that’s smaller, has a GPU, and is fabbed on a high-yield process.

    • Bensam123
    • 11 years ago

    AM socket envy.

    • Bensam123
    • 11 years ago

    I don’t think anyones disputing BD’s having half as many real cores vs logical ones…

    • maxxcool
    • 11 years ago

    [url<]https://techreport.com/articles.x/21987/13[/url<]\ a cpu limited to 4 cores per cycle = 4 cores, not 8.

    • Bensam123
    • 11 years ago

    That’s great and everything if you believe AMD or Intel marketing, but people on this site are smart enough to know the difference. Chances are, based on Intels market share, people don’t believe their marketing either. Both companies spin marketing in a way that makes their company look best, they’ve been doing it forever.

    MS released the performance patch for BD in Win7, TR did a report on the results from. It didn’t make a huge difference, but that was just an addnum, not the point of what I was discussing.

    Keep in mind your 6 core i7 is a $1000 CPU and a 8 core BD is $200. I’d personally like to see what the 12/16 core models can do against Intels CPUs, since AMD server CPUs are so cheap they start falling into the consumer bracket.

    • maxxcool
    • 11 years ago

    Mark me down all you want, that 8 core cpu cannot even beat a 6 core I-series cpu…

    sarcasm aside, realtiy check. It may have 8 “cores”. but it cannot use more then 1 core per module at one time as revealed here on techreport and andandtech.

    if you cant use more than 1 core per module… thats a 4 core cpu.

    [url<]https://techreport.com/articles.x/21987/13[/url<]

    • NeelyCam
    • 11 years ago

    [quote<]Unless you're being paid a pittance, most people wouldn't care about a measly 4% difference.[/quote<] It's a 20% difference. Or 25%, depending on if you're in a 'glass half-full' or 'glass half-empty' kind of mood

    • NeelyCam
    • 11 years ago

    … but how saving money for an IB/SB build would help if you’re going to sleep until Haswell anyway..?

    • maxxcool
    • 11 years ago

    Yet their marketing keeps rolling out the banner ad’s calling them 8 cores cpu’s when a 6 core I-series would wipe the floor with it.

    Windows-8’s new scheduler wont save this cpu either. Expect very similar lackluster perfromance.

    • just brew it!
    • 11 years ago

    If you’re overclocking you’re probably not too concerned with CnQ anyway…

    • Deanjo
    • 11 years ago

    It may not be but it is not guaranteed as it is with higher end models and you are still sacrificing proper functioning CnQ to overclock.

    • maxxcool
    • 11 years ago

    Only for 4 cores active at 1 time = 4 cores. The core scheduler does not allow for more than 1 core per module to perform a operation as shown on this very site in review. So if you can’t use more than 1 core per module, you have only 4 cores active at one time. which is why a “8” core 8150 cannot beat a 4 core I-series by any significant margins.

    • ish718
    • 11 years ago

    8 int cores count as 8 cores.
    One FPU per module with slow cache sucks though.

    • Bensam123
    • 11 years ago

    If I had to make a guestimate I would say they overpriced them for the first couple months to get more money out of early adopters and people who would be willing to pay extra for the name, before they go back to their old practices, which were quite successful. I don’t believe AMD is stupid enough to continue the practice till they run themselves into the ground.

    There will probably be another round of price cuts in a couple months again bringing them down to competitive price/performance or outright better as they’re the underdog and have to do what they can to get their sales.

    • Bensam123
    • 11 years ago

    If what I read was right from someone else who posted links to benchmarks on another website (forgot which site it was), AMDs version of hyperthreading is actually very well done and out performs Intels implementation by roughly 33-50% in a lot of cases.

    This is of course cases where virtual threads stand out, not talking about four actual cores vs their four make-believe cores and as long as the system has the patch that makes it recognize AMDs hyperthreaded modules as that and not actual cores. Essentially hyperthreading vs AMDs second virtual core.

    Just people don’t care as cases where this is relevant are slim to non-existent.

    • Vivaldi
    • 11 years ago

    Nailed it! +1

    • Bensam123
    • 11 years ago

    If AMD can go back to matching price/performance ratios people would be willing to buy their processors. Overpricing a product that doesn’t perform nearly as well is stupid.

    • just brew it!
    • 11 years ago

    One more price cut like this and I’ll consider buying an FX-8xxx CPU!

    • flip-mode
    • 11 years ago

    When is that rumored to happen?

    • maxxcool
    • 11 years ago

    Agree. This fake 8 core BS can suck it.

    • just brew it!
    • 11 years ago

    I have [i<]never[/i<] heard of percentage difference being calculated based on the average. It is supposed to be based on the percentage change [i<]compared to the original value[/i<].

    • flip-mode
    • 11 years ago

    Hmm… maybe… I’d still rather put the money toward an Ivy / Sandy build…

    • flip-mode
    • 11 years ago

    As was said toward the end of that post, I’m not after speed; I’m after capacity. I have 8 GB but could use more for VMs. I could buy 4 GB DDR2 DIMMs, but have you seen how much they cost? 16 GB of DDR2 would cost me a minimum of $300. I could buy an AM3+ mobo and 32 GB of DDR3 for less than that.

    • maxxcool
    • 11 years ago

    4 cores, not 8. 4 + very poorly designed amd-hyperthreading.

    • kamikaziechameleon
    • 11 years ago

    That is true, but if you didn’t want to OC then why not get a better stock performance part(like an intel quad)?

    If I had to find value in the FX line it would be with the 8120 OC’d nicely and put to task in Multi threaded applications. @ 165 it doesn’t really compete with the 2600k or any 8 threaded intel processors or even the single threaded quad lineup. Meanwhile for the 200 dollar price point the 8150 has no appeal as intel beats it hands down. Honestly I don’t see value in anything AMD offers over the 160 price point as simply put they don’t compete.

    • kamikaziechameleon
    • 11 years ago

    ok yeah that is a true statement! Its funny how Core i killing the FSB really changed the game.

    • kamikaziechameleon
    • 11 years ago

    But from looking at the OC reviews it wasn’t heat constrained during OCing if I’m not mistaken so that may not be a concern.

    • Deanjo
    • 11 years ago

    [quote<]My 955 BE is doing fine too and is certainly not looking bad vis-a-vis BD, but I'm on DDR2 and feeling the limits there quite acutely. [/quote<] I very much doubt that. The effects of DDR2 vs DDR3 in real life usage on the AMD processors is next to nothing.

    • Deanjo
    • 11 years ago

    The only thing that I have encountered where intel was consistently better was on their NIC’s (as long as the board wasn’t crippled with a 3rd party solution).

    Until recently with the introduction of the SB chipsets, AMD had the upper hand in bus bandwidth and connectivity support.

    • Sam125
    • 11 years ago

    Unless you’re being paid a pittance, most people wouldn’t care about a measly 4% difference. Unless you’re the taxman of course. ; )

    Of course it’s always best to be accurate but I’d file this scandal under “who cares” just below “it might be interesting if I were a lone eskimo living on the South Pole”. : P

    • hansmuff
    • 11 years ago

    And the real shame is that the A75 chipset on the FM1 boards is really, really nice.

    I put a FM1 box together as a “family PC” for everything from Minecraft over homework to simple video processing. This machine has not ever bluescreened, given me driver problems or otherwise griefed me. Really a wonderful build.

    I knew FM1 was going to be short lived, went for it as I tend to buy entire platforms anyway.

    • sschaem
    • 11 years ago

    Denial ?

    Again what is 20% of 205 ? 41
    Add 41 to 205.

    205 + 20% is .. tada

    246.

    How dense can you be ?

    Its fine to ‘lol’ and be cocky all you want… but you better be right or you look like a fool.

    (Hope I didn’t break any rules by saying it like it is, I’m surprised the s word is taboo)

    • Bensam123
    • 11 years ago

    Marked down because I’d still have motherboard and memory that would work with everything up to the Phenom 2 x6?

    • kamikaziechameleon
    • 11 years ago

    hope you didn’t get burned. Perhaps my experiences are limited in scope but I’ve had the opposite experience.

    • xeridea
    • 11 years ago

    The 2500k isn’t significantly faster in most things, and is a good bit slower in multithreaded tests, you know, what the 8150 was made for. The 8150 can match or beat the 2600k in some multithreaded tests. Also you can’t just ignore the GPU in Llano, just because you know it totally crushes the competition. Also, in general, Intel motherboards cost more and have less features.

    • yogibbear
    • 11 years ago

    Yeah I’ll give you a 20% pay rise… I mean 16%…. but whatever…

    • Duck
    • 11 years ago

    You exaggerate. Margins have been squeezed. It happens. Dual module bulldozer should be less than 160mm². Sandy bridge is 131mm²/149mm² depending on GPU.

    • Chrispy_
    • 11 years ago

    They’ll need a bigger price cut than this to clear inventory, the FX-8150 is still priced within $15 of a 2500K which is [b<]significantly faster[/b<] than the AMD for almost everything you can test for. The A8's are still quite appealing, but if you ignore the IGP in a Pentium G620, it is a faster, more efficient CPU than any Llano for only $60. It's not hard to find a GT430 deal for $45, bringing the total to $105. Unfortunately for AMD, the GT430 is (at worst) a match, and in many cases, [i<]totally obliterates[/i<] Llano's IGP.

    • glacius555
    • 11 years ago

    Indeed, who cares, 16% or 20%? We can afford it anyway being 1st world, right? /sarcasm

    • NeelyCam
    • 11 years ago

    [quote<]it's not a complete and unmitigated disaster that some people make it out to be.[/quote<] Well, no - not for the customer. But considering the die size, it [i<]is[/i<] an unmitigated disaster to AMD.

    • flip-mode
    • 11 years ago

    My 955 BE is doing fine too and is certainly not looking bad vis-a-vis BD, but I’m on DDR2 and feeling the limits there quite acutely. So for me it’s a question of what to do to get to DDR3 – buy an AM3+ mobo and keep riding the 955? That’s certainly an option; it’s not at all exciting but it would save me a fair bit of money. Or should I wait for Haswell? That’s also possible; my current rig can hold for another year since I don’t bump into my 8 GB RAM limit too often – only when messing with several VMs at once.

    I’d also like to get a genuine RAID card.

    • NeelyCam
    • 11 years ago

    [quote<]First world problem much?[/quote<] LOL well said. +1

    • NeelyCam
    • 11 years ago

    Are you sure you wanna pay for the mobo, though?

    • Bensam123
    • 11 years ago

    The AM socket is black magic… sorta makes me wish I bought one in the AMD 64 days.

    • Damage
    • 11 years ago

    Work around the profanity filter again, and you get banned.

    • Bensam123
    • 11 years ago

    Such is life around TR… Pointing it out is one thing, claiming the world the apocalypse is nye and saying what a terrible person Cyril is because of it is another.

    There are those that are OCD to correctness no matter the level.

    • Deanjo
    • 11 years ago

    [quote<]It is also the fact that the AMD mobo chipsets frequently struggle to compare to intel offerings.[/quote<] I just about spit up my coffee while reading that. I have come across more showstopping errata on intel chipsets then I have on any other (exception being via). How many SATA 3 ports do intel chipsets feature again?

    • Deanjo
    • 11 years ago

    Yes but you have to also remember that the TDP on both are the same at their respective rated speeds. Start overclocking 8120 and you may exceed that TDP of the 8150 just matching clocks. Plus overclocking can effect the power saving schemes leading to far greater amount of power consumption.

    • ish718
    • 11 years ago

    Some people do not overclock, and clock speed is very important in Bulldozer’s performance.

    • kamikaziechameleon
    • 11 years ago

    @ 165 you can get a nearly identical product with an unlocked upper multiplier!

    • kamikaziechameleon
    • 11 years ago

    what is the diff between the 8150 and 8120? All I see is clock speeds and looking at the “black edition” on the box tells me that clock speeds don’t really matter.

    • ImSpartacus
    • 11 years ago

    Lol, you’re one of those people I’m talking about.

    • BobbinThreadbare
    • 11 years ago

    I’m not going to do anything with 8 cores. I’d prefer a higher clocked 4 or 6 CPU. Also, I’d prefer a TR reveiw.

    • ish718
    • 11 years ago

    8 cores for $205, not bad.
    Too bad Newegg wants to charge $220 for the FX-8150.
    I noticed the ratings went from 4 stars to 5 stars on Newegg.

    Cheapest price I found is $210 at Frys

    • anotherengineer
    • 11 years ago

    Meh my 256GB Plextor SSD just came in today, it will probably give my AMD 955 BE at least another 3-5yrs of life.

    A lot cheaper than an entire new system too.

    • JoshMST
    • 11 years ago

    Yeah, it does seem that Piledriver does fix a lot of the thermal issues, and a small jump in IPC. Not a real big jump though… Vishera will be the real litmus test.

    • anotherengineer
    • 11 years ago

    Well how about $90 bucks

    [url<]http://www.ncix.ca/products/index.php?sku=70935&vpn=HDX945WFGMBOX&manufacture=AMD&promoid=1058[/url<] edit [url<]http://www.hookbag.ca/product/H3C06MNI2/[/url<] hmm maybe I should tell the old man it's time to upgrade his s939 single core 2.4 ghz a64, and the radeon x800gto, and 120gb ide hdd, but he says its fast and works good.

    • sircharles32
    • 11 years ago

    My apologies, I didn’t notice the Windows Media Encoder graphs until after you prompted me of my error.

    Basically, it seems that in order to get a small bump over the stock 1100T (at least with the WME tests), you’ve got to pay for it at the wall socket.

    I guess my fingers are crossed that they’ve been able to fix some of the electrical leakage problems with Piledriver.

    • JoshMST
    • 11 years ago

    I did test WME 64. Not the greatest encoder, but at least it is multi-thread aware. Plus consistent results…

    • colinstu
    • 11 years ago

    read the title as “AMD quietly cuts FX-, A-series product lines” at first.

    That would’ve been a bit dramatic.

    • sircharles32
    • 11 years ago

    Alright, that was pretty painful.
    92 Watt jump, and still falling behind 1100T.
    Even though the article didn’t use any video encoding, I don’t think I’d be inclined to run any long batch jobs with such a power drain.

    Thanks for the link.

    • JoshMST
    • 11 years ago

    This actually answers a lot of your questions: [url<]http://www.pcper.com/reviews/Processors/AMD-FX-6200-CPU-Review-Small-Bulldozer-Refresh[/url<]

    • Duck
    • 11 years ago

    No. I was elaborating on why I would be interested in seeing the smaller brethren tested. Which is what you were talking about too.

    • sircharles32
    • 11 years ago

    Tom’s already did a “gaming” review where they showed that at normal gaming resolutions (ie. 1920 x 1080), the CPU plays second fiddle to the video card anyway. That’s nice, but I don’t game. I’d be more interested in seeing mainly video encoding tests. Seeing as how I refuse to pay over $150 for a CPU, that pretty much knocks the quad module CPU’s out. That’s why I’m interested in the lesser models.

    • yogibbear
    • 11 years ago

    Did AMD send you that table? Or did you dump those no’s into excel after 5pm with no coffee?

    • flip-mode
    • 11 years ago

    Yes, I saw your link above and +’ed you.

    • ClickClick5
    • 11 years ago

    Is today THIS dull that you all have to bicker about the concept of a price difference?

    First world problem much?

    • Duck
    • 11 years ago

    Percentage change = ( ΔV / |V1| ) * 100 = ((V2 – V1) / |V1|) * 100

    • flip-mode
    • 11 years ago

    Had to thumb you down because your link says: [quote<] when there is no direction of change; no old and new value[/quote<]

    • sircharles32
    • 11 years ago

    I never said it was a “dog”.
    I also never said it was a “disaster”.

    Did you mean to respond to someone else?

    • flip-mode
    • 11 years ago

    (Initial Value) / (New Value) = Difference so 245 / 205 = 1.195 = 20% difference.

    Math never lies but you can make it tell different truths.

    • Duck
    • 11 years ago

    You need this [url<]http://www.calculatorsoup.com/calculators/algebra/percentagechange.php[/url<]

    • flip-mode
    • 11 years ago

    Wake me when AMD is [i<]giving FX away[/i<] and I'll take one. Otherwise, wake me for Haswell.

    • ronch
    • 11 years ago

    I think someone needs to look at why AMD is ditching FM1 so soon. It’s like a slap in the face for those who bought FM1 boards.

    • sschaem
    • 11 years ago

    What a load of ****…

    The table say “Difference”

    205 + 20% = 246 << this is correct (205 * 1.2) Go, go find a calculator and check.

    Go back to school.. please.

    • Anonymous Gerbll
    • 11 years ago

    Just for you duck.

    [url<]http://www.calculatorsoup.com/calculators/algebra/percentagedifference.php[/url<]

    • ronch
    • 11 years ago

    Man, if that’s the right way to do it, it doesn’t make sense to me.

    • Unknown-Error
    • 11 years ago

    It won’t be a game changer, but it won’t suck like Bulldozer either. It’ll be decent/respectable. Phenom sucked, but Phenom-II was respectable. Actually, mobile Trinity could to do pretty well in the market. Steamroller on the other hand, could be a major game-changer.

    • ronch
    • 11 years ago

    With regards to the current 4-cores and 6-cores, with Trinity and Vishera just around the corner I don’t think it would be of much interest to many gerbils here apart from being a curiosity. Not to be cruel, but I think today’s <8-core chips are just gonna fade into oblivion.

    Tom’s Hardware has recently somewhat included the FX-4100 in its monthly Best Gaming CPUs list, so maybe it has something going for it.

    • Duck
    • 11 years ago

    lol wut

    • dpaus
    • 11 years ago

    I’d like to see a comparison test of new Ivy CPUs and the AMD CPUs, culiminating in the famous-and-should-be-TR-patented ‘platform-cost value scatter plot’

    And you know what I think would be really cool? (hint, hint, Cyril): keeping the performance values in a table that gets updated, and the scatter-plot regenerated, everytime one of these price cuts happens. Even better, show the last two generations of value for each processor, so we can see how Intel and AMD are positioning their offerings over time.

    • Anonymous Gerbll
    • 11 years ago

    To calculate the % difference you need to divide the difference by the average. The difference is 40 and the average is 225. So: (40/225)x100 = 17.77% Neither 20% nor 16.3% is correct.

    • Duck
    • 11 years ago

    The FX-4100 costs almost exactly the same as the i3 2100. That’s locked dual core verses an unlocked quad core.

    I know bulldozer is a bit of a dog, but it’s not a complete and unmitigated disaster that some people make it out to be.

    • ronch
    • 11 years ago

    Alexko’s reply above hits the nail right on the head. The wording on the article doesn’t align with the table.

    • ronch
    • 11 years ago

    The price cuts were long overdue alright, particularly for the FX-8150. But come to think of it, it should have been $205 right at introduction. There shouldn’t have been a need for a price cut at all for you to be able to buy it for $205.

    • sircharles32
    • 11 years ago

    Not only would I be interested in seeing the smaller brethren tested, but I’d also like to see how it’s effected during over clocks. We’ve already seen that a quad module will consume about 300+ watts when clocked at around 4.8GHz. I’m curious how it’s effected when only 2 (41xx) or 3 (6xxx) modules are over clocked. Will it be significantly less (wattage)? How much farther can it be pushed (normal cooling, not liquid nitrogen/helium)? Most importantly, how will it perform, while under these conditions? That’s what I’d like to know.

    • Duck
    • 11 years ago

    But “20% off” translates into a -20% difference, or a +20% reduction.

    • ronch
    • 11 years ago

    Except that isn’t the norm. When stores say 20% off, 245 becomes 196.

    • Duck
    • 11 years ago

    Yes you are right there.

    • Alexko
    • 11 years ago

    This occurred to me too, but then the text relies on the table to say things like this:

    “The hardest hit were the FX-8150, A8-3850, A6-3650, A4-3400, and A4-3300, all of which got their prices slashed by 20% or more.”

    And this is clearly wrong, even though the table technically isn’t. It’s not a big deal, just a little misleading.

    • Duck
    • 11 years ago

    First +1 I’ve given you ^_^

    • ermo
    • 11 years ago

    Any reliable word on how the Trinity APUs will compare to the FX CPUs in terms of performance per watt, clock scaling and IPC?

    • ronch
    • 11 years ago

    I was supposed to point out the exact same thing. Then I saw your post.

    • Duck
    • 11 years ago

    Got to be super picky here…

    The column is labeled “Difference”, not “Reduction”. So for the FX-8150 BE, the difference is not marked as -20%, but 20% (i.e. +20%). Because of this, you could argue the table is correct and the “Difference” is the % increase of the new price to get to the old price.

    • glacius555
    • 11 years ago

    This is embarrassing…

    • Farting Bob
    • 11 years ago

    +42.

    • BobbinThreadbare
    • 11 years ago

    “Oh, AMD. What would you do without your die-hard fanboys?”

    Keeping selling integrated solutions to OEMs?

    • BobbinThreadbare
    • 11 years ago

    I think you might be overreacting to a mistake. It’s not like Cyril has a history of doing this.

    • ImSpartacus
    • 11 years ago

    Yeah, that’s pretty tacky. It’s amazing how people can go through years of schooling and still struggle with the concept of manipulating a fraction of a whole. I can understand typos, but this isn’t a typo.

    • Unknown-Error
    • 11 years ago

    No price drop for FX-4170?

    So much for that… :\

    • kamikaziechameleon
    • 11 years ago

    I doubt it. Seems even when they have a good idea the different divisions of the CPU portion of the company are always railing against one another. Its not that Bulldozer doesn’t have some ingenious ideas(it really does), its that it simply isn’t a coherent product.

    • ALiLPinkMonster
    • 11 years ago

    Oh, AMD. What would you do without your die-hard fanboys?

    I really hope Trinity is a game changer.

    • kamikaziechameleon
    • 11 years ago

    Its not just the issue with the CPU price performance not being remotely competitive. It is also the fact that the AMD mobo chipsets frequently struggle to compare to intel offerings. Add in the fact that their black editions are terribly relevant since they don’t OC very well and there is no reason to get an AMD product, not even if you already have a compatible mobo with a 1090T or something in it.

    • kamikaziechameleon
    • 11 years ago

    Still 20-40 dollars overpriced in many cases. This is the worse position AMD has been in with regards to consumer processors since the original botched launch of the phenom line.

    • Alexko
    • 11 years ago

    The percentages in the table are all computed backwards.

    The FX-8150, for instance, actually got a 100 – (205/245)×100 = 16.3% price drop.

    • BobbinThreadbare
    • 11 years ago

    Why buy such an expensive CPU for that? With the IGP handling decoding, I’d go with probably an A4.

    Unless you’re using it to encode recordings as well.

    • JustAnEngineer
    • 11 years ago

    +1.

    • Goty
    • 11 years ago

    I might be looking to pick up an A8-3800 to replace the X3 720 + 4670 combination in my HTPC if I can find a good deal on a mobo, too. I’d get basically the same performance of both the CPU and GPU in a package with a TDP lower than the CPU I’m running now.

    • jdaven
    • 11 years ago

    Price cuts for the A-series can possibly be explained by the imminent release of Trinity. They might be clearing inventory.

    • MadManOriginal
    • 11 years ago

    The A3850 isn’t a terrible CPU for ~$100 if it’s going to be used in a system without a discrete graphics card. Too bad it’s socket FM1 which is already set to be replaced, for non-APU CPUs AMD has always been good at socket longevity.

    • BobbinThreadbare
    • 11 years ago

    Is there going to be a TR review of other FX chips besides the 8100s?

Pin It on Pinterest

Share This

Share this post with your friends!