TR Forums

.

flip-mode wrote:

Opinions vary; mobos are pretty interesting things, IMO.

To be fair they can be equally frustrating/annoying, especially when you have to diagnose random instability which can be caused by random mobo component. I personally try to not upgrade mobos unless it is necessary, I'd rather expand their capabilities with, well, expansion cards

just brew it! wrote:

Geonerd wrote:

Used 1090Ts are going for $130 ~ $160.

Wow, those have held their value quite well for a CPU chip; that's close to what I paid for a new one around 2 years ago. I guess there must still be demand for these from people trying to squeeze the last bit of performance out of their aging AM2+/AM3 systems.

I'm one of those people. I keep hoping one will show up on Amazon since I don't use fleabay.
[Edit] Actually, I meant the 1100T, scarce as hens teeth. [/Edit]

JohnC wrote:

I personally try to not upgrade mobos unless it is necessary, I'd rather expand their capabilities with, well, expansion cards

I try not to upgrade anything unless it is necessary If only there was an expansion card that let me put a Haswell in my AM2+ motherboard. At that point the motherboard itself is the expansion card.

.

flip-mode wrote:

JohnC wrote:

I personally try to not upgrade mobos unless it is necessary, I'd rather expand their capabilities with, well, expansion cards

I try not to upgrade anything unless it is necessary If only there was an expansion card that let me put a Haswell in my AM2+ motherboard. At that point the motherboard itself is the expansion card.

Remember these?

.

No. This is an expansion motherboard for ECS PF88 mobo. There were few of these expansion mobos, each for different type of CPU (Pentium-M, Socket754 and Socket939). Back then both hardware and software companies actually cared about giving consumers as much choice as possible (even if these choices were largely impractical and unused).

JohnC wrote:

No. This is an expansion motherboard for ECS PF88 mobo. There were few of these expansion mobos, each for different type of CPU (Pentium-M, Socket754 and Socket939). Back then both hardware and software companies actually cared about giving consumers as much choice as possible (even if these choices were largely impractical and unused).

I think that last sentence is supposed to read, "Back then the market would support aftermarket solutions like this, instead of effectively forcing people to dole out the dosh needed to buy a brand new computer when the old one got creaky." I miss those days too, and if there's a single bright spot to the slow dwindling of the desktop it's that catering to expandability and upgrades may come back a bit. Prices will go up too, but the market will prevail.

clone wrote:

How is it not inconsistent to say "get the cheap cpu" and also "don't get a cheap mobo"?
the FX 8320 is viewed by many enthusiasts as "the cheapest toilet available".

because $145 to $500 vs $100 to $130 is not the same comparison..... $30 difference, meh, $200 + significant and steals from the overall budget.

You keep saying $500. It's $350. And, again, the additional money is buying additional hardware and additional performance - not just cpu performance - USB and SATA performance.

I must say that I never have real world experience with a mobo until I buy one, but I definitely can read specs and reviews and form my own opinions.

Flip Mode stating an uncomfortable fact, is not a personal attack and that doesn't mean I'm trying to offend anyone.

It's irrelevant. I can drop some components into a mobo and hit the power button on a computer and suddenly claim to have "experience" using this or that motherboard, or I can read a review that meticulously examines the motherboard and benchmarks it and evaluates the drivers and BIOS and in the end I'm much more informed by having read the review. You, however, are implying that you - some guy on the internet - can tell me better than a proper review, and I reject that.

saying "AMD demands too many compromises", "AMD's motherboards are so bad the worst Intel is exciting in comparison"... (paraphrasing) these conclusions you profess aren't real

They're opinions, and as such, they are every bit as real as yours. The fact that you think your opinions are more real than other people's is... a problem.

you've already mentioned that your X4 AMD cpu/mobo is handling everything you ask and the new AMD CPU's/mobo's are superior yet you are applying terms like "compromise" and almost hinting that AMD's SATA performance is broken.

AMD's SATA and USB performance is not as good as Intel's. Cpu performance is not as good as Intels. Power consumption is not as good as Intel's. So, regardless of what I have (being that information pertaining to the hardware I currently have is quite irrelevant to what my opinion is regarding the purchase of hardware released subsequent to 2009), if I was buying today, I'd buy a Haswell 4670k.

I'm not saying AMD is perfect, I'm not saying Intel is perfect, I'm saying never forget the money and for heavens sake keep it in context.

That's great. That's your opinion and your advice. I don't need to invalidate it; there's no reason for me to do so. It's not like it's absurd or clearly poor advice. It's a fine opinion you have there. I just don't feel the same. My opinion is that a FX 8320 is not a good upgrade from a 1090T at 4.1 GHz. My first thought is sit on the 1090T for a bit longer. My second thought is that if an upgrade is nonetheless still desired, then the FX 8320 isn't a significant enough upgrade, and that upgrading to Haswell gives demonstrably better CPU, SATA, USB and power consumption performance. Just like your opinion, my opinion is valid and it's not absurd or clearly poor advice.

To simplify: In my opinion, do not get the FX 8320, stick with the 1090T or else upgrade to Haswell i5-4670k for $350 or so.

But one thing..., the OP mentioned the 8230, not even the 8350. And thats only $35 difference. so clearly this was a cost conscious decision.
In the end, yes, the OP could spend over $350, and rebuild his system to get better performance. But clearly that was not the goal.

Yet, from what I can see, going from a 4ghz 1090T to a 4ghz FX-8320 is just not worth it.

sschaem wrote:

But one thing..., the OP mentioned the 8230, not even the 8350. And thats only $35 difference. so clearly this was a cost conscious decision.
In the end, yes, the OP could spend over $350, and rebuild his system to get better performance. But clearly that was not the goal.

Yet, from what I can see, going from a 4ghz 1090T to a 4ghz FX-8320 is just not worth it.

Agreed. In my mind I keep thinking better to spend $350 on a "real" upgrade than $150 on a marginal upgrade. But, as you said, that's not a direct answer the OP's question. He asked for more of a yes/no on the 8320, to which I agree with you: no.

As far as I'm concerned, IPC still matters and Bulldozer and Piledriver haven't actually surpassed the older Phenom II architecture in that respect.

Given that the 1090T runs at 4.1GHz, I'd be tempted to say that for gaming, the FX-8320 may very well be a downgrade!

It's tempting to think that way, but ultimately it boils down to power consumption and performance. An FX 8320 will clock higher than a 1090t ever will so without googling to try to find a direct comparison I can't comfortably guess which would come out faster. Judging from TR's reviews, the FX 8320 will actually draw less power (according to the FX 8350 review).

8350 power consumption overclocked: 262 watts
http://techreport.com/review/23750/amd- ... eviewed/13

1090t power consumption overclocked: 278 watts
http://techreport.com/review/18799/amd- ... cessors/14

flip-mode wrote:

8350 power consumption overclocked: 262 watts
http://techreport.com/review/23750/amd- ... eviewed/13

1090t power consumption overclocked: 278 watts
http://techreport.com/review/18799/amd- ... cessors/14

Note that the test system is drastically different between those two reviews...a GTX 260 and HD 7950 don't even come close to using the same amount of power at idle let alone the rest of the parts (a 260 idles at at least ~20-25 watts more than a 7950).

That said, I'd stick with the 1090T at well.

Wow, unbelievable amount of responses! Sorry I was out of town all weekend, but i did order the chip before I left. I'm going to have to go back and read it all.

I am definitely going to sell the 1090t and if I can get $100+ for it that would be awesome. $50 for a bit better performance and a whole lot of fun and experience with overclocking these newer chips is what does it for me. I know 5Ghz is going to be pushing it, but as long as I get at least 4.6/4.7 I'll live. I know about all the core parking stuff so I will definitely be messing with that.

As far as overclocking in windows or messing with power states etc I'll have to look into it. While it would be nice for a setup with auto overclocking/turbo or whatever it's called these days, I'd rather have excellent system stability and better performance. I have been overclocking for a while and I know it takes some time but it the end it's worth it to me.

As far as the guys wanting me to go Intel, that would be nice and all, but I don't have the money to spend on a whole new setup, nor do I want to. In fact I am perfectly happy with the 1090t I already have, but I'm bored and didn't want to pass on a nice deal. Might be upgrading the GPU in a similar fashion as well with all the deals lately.

For those wondering I am on a custom ghetto water cooling setup, basically two fat 120mm radiators outside of the case using an aquarium pump and generic cpu block/gpu blocks in the same loop. It has worked really well for many years and is incredibly quiet. (Actually had the 1090 @ 4.3Ghz but it was not perfectly stable, so I dropped to ~4.1/1.425v, GPU is a 6850 resistor modded to 1.35v @ 1050/1150)

madmanmarz wrote:

I did order the chip before I left. I'm going to have to go back and read it all.

Congratulations on your new hardware. You probably don't want to read the rest of the thread.

Given what I just saw on ebay you might nearly break even.

JustAnEngineer wrote:

madmanmarz wrote:

I did order the chip before I left. I'm going to have to go back and read it all.

Congratulations on your new hardware. You probably don't want to read the rest of the thread.

LOL I stopped reading when I saw deleted posts and drama but I might look it over when time allows

flip-mode wrote:

Given what I just saw on ebay you might nearly break even.

Sweet

I will post my results later this week with the overclocking as well =)

.

Yup, threads like these degrade whenever you suggest AMD as a good alternative around here, funnily so when it comes to budget conscious setups (which AMD definitely excels at). You should post back what sort of OC you get out of it. Overclockers.net has a OC club for 8320/8350 users as well.

JohnC wrote:

Bensam123 wrote:

I've played 64 person metros WHILE STREAMING and get 60-100fps, usually around 75ish.

The results depend on many different things, dude. Such as different in-game level - for example, my "average" on 64-player Operation Metro is 115FPS (according to FRAPS benchmark which I ran during whole gaming round), but on 64-player "Ziba Tower" it is around 100FPS ("Ziba Tower" is a CQ map with more "destructible" environments). Also depends on a video card. So you cannot compare your own results to the results of the benchmark unless you have same exact (or very close to it) hardware and play on same exact map with same exact number of people

Also, on your stream I've seen your FPS drop to less than 50. Which is close to min FPS of that Russian site's benchmark

Minimum FPS is not the same as average FPS. I've played quite a few other games besides BF3, if that was indeed in relation to watching me play BF3 (which I sorta doubt). You seem to be one of my viewers I have not had the pleasure of talking with. However, it is entirely possible to see my FPS drop to 50 at some point during BF3. But that's not 99% of the time. I also get more then 100FPS at times, but I excluded that.

And yes, it highly depends on the scenario. The point I was making was that he was making it seem like the 8350 simply is a bad option in BF3 and 'can't hold it's own', but that depends a lot on the map and other things (hence my statement about my dad beating up his dad). It's even possible for it to do just as well as new Intel hotness everyone is drooling over (blasphemy I know).

Yup, it may not be directly comparable because in the benchmarks I posted they were using a 7950, where as I am using a 7870, so it's entirely possible to get better FPS then what I posted.

Bensam123 wrote:

Yup, threads like these degrade whenever you suggest AMD as a good alternative around here

Neither can you expect to go unharassed if you politely recommend an Intel setup.

Guys (everyone), I think we already have all the pros and cons listed here, can we continue on track please?

anecdotal, but running a raytracing test I see the same power usage betwen a Q6600 at 3.2ghz and this fx-8320 at 3.5ghz . in this case ~95w above iddle.
~10.2 fps FX-8320 vs ~6.6fps Q6600 (Yes, my main system is still a Q6600)

I will have to play more with this system, but not to much because I just got an FF motherboard deadlock and had to reset the cmos...
(wonder why, because the only thing I tried was to set the VDD at 1.3 and the voltage at 1.26)

Soo.. what did I do wrong ? I'm now afraid to play with both setting. So right now I using a 100% stock setup but:

a) disabled turbo
b) set cpu vdd to 1.1125

I'd hold on to that OC'd 1090T if I were you. The FX-8320 may be a bit faster for most things but I'm not sure it's gonna be worth the upgrade.

Here's a little reference you might find handy.

sschaem wrote:

Soo.. what did I do wrong ? I'm now afraid to play with both setting. So right now I using a 100% stock setup but:

Well it sounds like you were changing voltages without doing any stability testing...

ronch wrote:

I'd hold on to that OC'd 1090T if I were you. The FX-8320 may be a bit faster for most things but I'm not sure it's gonna be worth the upgrade.

Here's a little reference you might find handy.

Unfortunately, those comparisons are all at stock. If OP manages to get the 8320 to 5 ghz it would be pretty great. That should not be counted on; I guess 4.6 to 4.7 is typically achievable on air. So what's the performance of 8320 at 4.7 compared to 1090 at 4.1? That gives the 8320 13% more hertz than the 1090. Interestingly, that's probably about the IPC advantage that the 1090 has over the 8320, so it's possible they could come out close to the same performance as long as thread count is not a factor. At higher thread counts, the 8320 will probably pull cleanly ahead in most cases.

Bensam123 wrote:

Planetside 2 is a steaming PoS when it comes to performance.
....
Oh, it also uses more then two threads if you watch the distribution of it in resource monitor or take the time to look at the thread count. There are almost no games that run on JUST two threads or less. They may not be evenly balanced, but they most definitely spin off into more, which Windows then distributes around the cores in whatever way it sees fit.

Yes it technically uses more than 2 threads, but in effect it really only uses 2. I have never seen my total CPU usage go above 50% when running PS2. Using half of 4 cores isn't really any better than fully using 2 cores.

I completely agree on it being a steaming PoS performance wise though.

Waco wrote:

sschaem wrote:

Soo.. what did I do wrong ? I'm now afraid to play with both setting. So right now I using a 100% stock setup but:

Well it sounds like you were changing voltages without doing any stability testing...

The voltage was within spec, I just wanted to set the VDD a bit higher then the cpu voltage as it seem its the way 'everyone' is doing it.
From a VDD of 1.26 to 1.3 and set the voltage to 1.26
So just setting the VDD at 1.26 : works
But doing VDD at 1.3 and voltage at 1.26 causes a bios lockup.

What I found is that this chip can compute all day long at a given voltage (prime95 small set), but will fail now and then doing L3 cache/memory access.
So the entire chip voltage need to be raised for just that reason, causing extra power draw. So what happen is that I'm overvolting the entire chip to get the L3 to behave.
It seem to me AMD got an easy way to get 10-20% better power efficiency if they can align their L3 cache with the x86/L1/L2 cache units. (based on my super limited hands on experience)

So all in all I wanted to see if the VDD was higher then the set voltage, this could help in the situation when power draw would go up.
Thats a small detail and I will play it safe for now and just set the VDD in the bios and be done.

The sweet spot seem to be 4ghz and I'm confident the machine will be silent loaded at that speed.
I'm actually happy enough that I will get a desktop case for this and make it my next dev system. But its really not a huge step up from a Q6600...
But then I'm really after SSE4/AVX so this is good enough. But I'm not sure I will keep this system as long as the Q6600 (going on 7 years now)

But for anyone with an AM3 / AM3+ motherboard this $145 chip is an amazing value, if you dont already have a 4~ghz phenom x4 or x6.

For the replacement server, dont need it just yet, I will go LGA2011 / i7-4930k for sure.

Alright guys best I could do was 4.4 Ghz @ 1.45v 2475 NB/HT, 1500 DDR3 7-8-7-1T. This thing puts out a TON of heat. Gonna have to do some little upgrades if I want moar gigaz. Sellin the 1090 but I guess if I had to do it all over again I wouldn't have upgraded without better cooling (Unless I DO get $150 for it).