Personal computing discussed

Moderators: renee, Flying Fox, morphine

 
geekl33tgamer
Graphmaster Gerbil
Topic Author
Posts: 1023
Joined: Tue Aug 25, 2009 7:25 pm
Location: England
Contact:

Not really an 8-Core CPU then?

Wed Apr 25, 2012 3:42 pm

Anyone out there with an AMD FX 8120 or 8150 to check this???

In the Windows 7 System Info Tool, and in CPU-Z my 8120 is reported as so:
Processor AMD FX(tm)-8120 Eight-Core Processor, 4400 Mhz, 4 Core(s), 8 Logical Processor(s)

My Media Centre / Home Server is also in the FX camp, and reports:
Processor AMD FX(tm)-4100 Quad-Core Processor, 3600 Mhz, 2 Core(s), 4 Logical Processor(s)

Say............what?
Mega Beast - Intel i7 4790K | Gigabyte Z97X-UD3H-BK | 32GB DDR3 | SLI GTX 1070 | Samsung 850 Evo 1TB
Mini Beast - Intel C2E QX9770 | Gigabyte X48T-DQ6 | 16GB DDR3 | KFA2 LP GTX 750Ti | Seagate 2TB SSHD
 
srg86
Gerbil Team Leader
Posts: 262
Joined: Tue Apr 25, 2006 7:57 am
Location: Madison, WI

Re: Not really an 8-Core CPU then?

Wed Apr 25, 2012 3:49 pm

afaik that processor branding string is actually written into the CPU by the BIOS/UEFI during POST, so I guess it's whoever wrote that code in there shows it at this, though I would have thought AMD's documentation (probably BIOS writers guide) would instruct what to string to program in.

EDIT: Scratch that, I assume the 4 cores and 8 logical processors thing is what CPU-Z and windows are saying.
Intel Core i7 4790K, Z97, 16GB RAM, 128GB m4 SSD, 480GB M500 SSD, 500GB WD Vel, Intel HD4600, Corsair HX650, Fedora x64.
Thinkpad T460p, Intel Core i5 6440HQ, 8GB RAM, 512GB SSD, Intel HD 530 IGP, Fedora x64, Win 10 x64.
 
tfp
Grand Gerbil Poohbah
Posts: 3413
Joined: Wed Sep 24, 2003 11:09 am

Re: Not really an 8-Core CPU then?

Wed Apr 25, 2012 4:04 pm

Depending who you ask AMD FX CPUs are not 8 full cores and AMD is just using a "better" version of hyperthreading. OSes can not treat each thread on those CPUs as a full process for best performance so MSFT must just those cores as logical processors. Until right before launch AMD itself didn't call each of the 8 "modules" cores but marketing change that to help fight the "i have the most cores in my CPU battle".
 
cmm05089
Gerbil In Training
Posts: 7
Joined: Sat Mar 31, 2012 10:23 am

Re: Not really an 8-Core CPU then?

Wed Apr 25, 2012 5:01 pm

http://techreport.com/articles.x/19514

4 modules, where each module includes 2 integer 'cores' but shares FPU, cache, and fetch/decode logic.
 
vargis14
Gerbil Jedi
Posts: 1900
Joined: Fri Aug 20, 2010 6:03 pm
Location: philly suburbs

Re: Not really an 8-Core CPU then?

Wed Apr 25, 2012 6:04 pm

It does kinda sorta:) have 8 cores broke into 4 modules.L1 cache size is broken into 4 x 64 KB two-way associative shared instruction caches along with 8 x 16 KB 4-way associative data caches
Level 2 cache size is 4 x 2 MB 16-way associative shared caches.The 8mb level 3 cache is strangely split into 4 parts.
In layman's words it has 4 modules broke into 2 cores per module unfortunately each module/2 cores has to share a single 64kb L1 cache and a single 2mb L2 cache then work in 2 16kb associative data caches which i presume help divide the work from the 2 cores in the module to the single L1 and L2 caches.The sharing of l1 and l2 caches along with the scheduling of processes has to greatly increase latency's to the individual 2 cores in the module making them more like hyperthreaded cores then true cores.In fact windows had to rewrite a good bit of code to get it to work like intels hyperthreading but the improvement was minimal to none.
Bottom line is for intel and amds phenoms to be a true 4 or 6 core cpu each core needs its own L1 and almost positive L2 caches to be considered a true core......not a module consisting of 2 wannabe cores sharing a single L1 and L2 cache.
Now as for L3 caches they are shared among all the cores to store instructions and such.But they are not needed they just improve cpu performance allowing a lot more commands to be stored right on the cpu die and not in system memory thus vastly improving latency for commands/instructions so the cpu does not have to wait to take action.
I hope i am making my self clear and above all else I am correct.I could be wrong here and there but if my memory serves me right i think i explained it well.
P.S. stay away from Bulldozer if poss,sandy and ivy bridge are way better cpus.I was a AMD fanboy untill i read the sandy bridge reviews and built a system for myself.AMD is still strong in my heart i have high hopes for trinity for laptops and HTPCs and also hope piledriver is not a bulldozer with a mesh tech band aid on it.If you are religious then pray for AMD if not keep your fingers crossed and hope AMD pulls not a wimpy rabbit out of there hat but a tiger or something :P
Last edited by vargis14 on Wed Apr 25, 2012 6:09 pm, edited 1 time in total.
2600k@4848mhz @1.4v CM Nepton40XL 16gb Ram 2x EVGA GTX770 4gb Classified cards in SLI@1280mhz Stock boost on a GAP67-UD4-B3, SBlaster Z powered by TX-850 PSU pushing a 34" LG 21/9 3440-1440 IPS panel. Pieced together 2.1 sound system
 
morphine
TR Staff
Posts: 11600
Joined: Fri Dec 27, 2002 8:51 pm
Location: Portugal (that's next to Spain)

Re: Not really an 8-Core CPU then?

Wed Apr 25, 2012 6:06 pm

Holy wall of text, Batman. Cat ripped out your Return key? :o
There is a fixed amount of intelligence on the planet, and the population keeps growing :(
 
vargis14
Gerbil Jedi
Posts: 1900
Joined: Fri Aug 20, 2010 6:03 pm
Location: philly suburbs

Re: Not really an 8-Core CPU then?

Wed Apr 25, 2012 8:00 pm

morphine wrote:
Holy wall of text, Batman. Cat ripped out your Return key? :o

Whoops i just started typing what i could remember as it flowed.
At least i used a few periods and commas.
Was never had great writing skills and they actually suck pretty bad.Its been 25 years since high school.Then add texting on phones and MSM messenger and alike programs for just about 15 years perhaps longer. I should hope everyone can understand what i am saying.

My posts definitely are not pretty,but i hope i get my point across and my info as accurate as i remember it.

I do have many fans that love my posting style LOL NOT!
2600k@4848mhz @1.4v CM Nepton40XL 16gb Ram 2x EVGA GTX770 4gb Classified cards in SLI@1280mhz Stock boost on a GAP67-UD4-B3, SBlaster Z powered by TX-850 PSU pushing a 34" LG 21/9 3440-1440 IPS panel. Pieced together 2.1 sound system
 
morphine
TR Staff
Posts: 11600
Joined: Fri Dec 27, 2002 8:51 pm
Location: Portugal (that's next to Spain)

Re: Not really an 8-Core CPU then?

Thu Apr 26, 2012 3:39 am

It's cool, that was a joke more than anything else. The post was informative nonetheless.
There is a fixed amount of intelligence on the planet, and the population keeps growing :(
 
srg86
Gerbil Team Leader
Posts: 262
Joined: Tue Apr 25, 2006 7:57 am
Location: Madison, WI

Re: Not really an 8-Core CPU then?

Thu Apr 26, 2012 7:36 am

tfp wrote:
Depending who you ask AMD FX CPUs are not 8 full cores and AMD is just using a "better" version of hyperthreading. OSes can not treat each thread on those CPUs as a full process for best performance so MSFT must just those cores as logical processors. Until right before launch AMD itself didn't call each of the 8 "modules" cores but marketing change that to help fight the "i have the most cores in my CPU battle".


Absolutely. I'd even go as far as saying AMD calling these an 8 core CPU is false advertising. There are 8 integer back ends but they share both 4 front ends and 4 (decent) FPUs. It's totally 4 cores with a better version of hyperthreading. It even behaves just like hyperthreading to the operating system. I don't think the Win 7 patch would have been needed if the CPU would tell the operating system that it had hyperthreading, which the intel cpus already do with the CPUID instruction.
Intel Core i7 4790K, Z97, 16GB RAM, 128GB m4 SSD, 480GB M500 SSD, 500GB WD Vel, Intel HD4600, Corsair HX650, Fedora x64.
Thinkpad T460p, Intel Core i5 6440HQ, 8GB RAM, 512GB SSD, Intel HD 530 IGP, Fedora x64, Win 10 x64.
 
Flatland_Spider
Graphmaster Gerbil
Posts: 1324
Joined: Mon Sep 13, 2004 8:33 pm

Re: Not really an 8-Core CPU then?

Thu Apr 26, 2012 10:42 am

From my understanding of the Windows patch, that would have solved the problem. There were some tweaks in the patch as well, but getting the OS to recognize that the procs have multithreading was the big thing.

I'm not sure AMD could do that due to trademarks and what not. Of course a better front end might have prevented this in the first place.

The Bulldozer procs are definitely multithreaded chips. Cut the marketing number in half, and that's how many real cores you have.
 
thefumigator
Gerbil In Training
Posts: 2
Joined: Fri Jul 29, 2011 12:05 pm

Re: Not really an 8-Core CPU then?

Thu Apr 26, 2012 11:02 am

To me each core in FX is hard to define.

However, looking at the core pictures you can spot 8 cores and also spot the shared resources.

If you see intel's core i7 snapshot what you see is 4 very defined cores, yet each core has a virtual extra core. but you don't spot any extra core actually.
 
geekl33tgamer
Graphmaster Gerbil
Topic Author
Posts: 1023
Joined: Tue Aug 25, 2009 7:25 pm
Location: England
Contact:

Re: Not really an 8-Core CPU then?

Thu Apr 26, 2012 3:35 pm

I feel kinda cheated a bit. I looked at the CPU die diagrams on how it's all held together - In theory it looks to be a better design choice than HT on Intel's CPU's. We all know in reality thats not the case, but there's definatly 8 core's (of sorts) that run at a nominal speed and will carry out a thread each for 8 in one go.

I don't get why the decided to use only 4 FPU's and share them with a pair of so-called core's. Logic dictates there should be 8, and if it had 8 performance would be pretty decent (again in theory). Is the FX 4100 1/2 of a FX 81xx then, so only has 2 FPU's, 4 so-called cores behind that and all of the bigger brothers cache at L3?

Would explain why the FX 4100 seems a little slow...
Mega Beast - Intel i7 4790K | Gigabyte Z97X-UD3H-BK | 32GB DDR3 | SLI GTX 1070 | Samsung 850 Evo 1TB
Mini Beast - Intel C2E QX9770 | Gigabyte X48T-DQ6 | 16GB DDR3 | KFA2 LP GTX 750Ti | Seagate 2TB SSHD
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Not really an 8-Core CPU then?

Thu Apr 26, 2012 4:47 pm

Given the shared resources, it probably makes sense for the OS to see it as a hyperthreaded processor for scheduling purposes. The reality is that it lies somewhere between "classic" Intel-style hyperthreading and fully independent cores.
Nostalgia isn't what it used to be.
 
ish718
Gerbil First Class
Posts: 189
Joined: Fri Feb 22, 2008 1:03 am

Re: Not really an 8-Core CPU then?

Thu Apr 26, 2012 5:02 pm

If all workloads were INT based then Bulldozer would be a beast. It was built to be part of a APU where the GPU handles FP work.
 
UberGerbil
Grand Admiral Gerbil
Posts: 10368
Joined: Thu Jun 19, 2003 3:11 pm

Re: Not really an 8-Core CPU then?

Thu Apr 26, 2012 5:14 pm

Certainly from any high-level view it's clearest to think of each module as a hyperthreaded core, but at the level the OS scheduler works it's not at all clear that treating it purely that way is the optimal thing to do. (Presumably the Windows patch would have been out earlier if it was.)

I guess this was all news to you, rcs2k4, and the marketing from AMD (and its partners) doesn't go out of its way to clarify things, but this issue got quite a bit of debate a couple of years ago when the Bulldozer design was first disclosed, long before the chips were actually available. (Here's a post I made on the topic a year ago, though you may find the whole thread interesting; the "Kanter article" I refer to is here).
rcs2k4 wrote:
I don't get why the decided to use only 4 FPU's and share them with a pair of so-called core's. Logic dictates there should be 8, and if it had 8 performance would be pretty decent (again in theory). Is the FX 4100 1/2 of a FX 81xx then, so only has 2 FPU's, 4 so-called cores behind that and all of the bigger brothers cache at L3?
It cuts down the die size, meaning more chips per wafer and less chance for a flaw per chip. That makes it cheaper / more efficient for AMD. For more background on AMD's logic, see page two of that Kanter article.

Sharing the FP unit makes sense for server chips, since a lot of server loads don't do any FP work at all: webservers just want as many integer cores as possible, for example, and many database loads are well-threaded but have minimal FP demands. There is precedent for this in server-oriented processors: Sun's first Niagara chip (UltraSPARC T1) went even further in this direction, sharing one FP unit per four cores (and 16 threads), or even more. (That was a little too extreme, however, even for Sun's server customers, and the T2 version added an FP unit per core, though shared by eight threads.)

Of course such a design is not well-suited to the demands of the "enthusiast"/workstation desktop and especially gaming, which tends to be FP-intensive, though we might note that the number of games that can keep even four cores busy with FP loads is still small (albeit growing). Bulldozer really does look like a server chip being sold in non-server markets: it doesn't have integrated graphics either, for example. Even the Trinity chip, which will include an IGP alongside an improved Bulldozer core, isn't going to be what you're looking for (its natural market is mobile, where the IGP should be adequate for gaming at typical laptop resolutions). But then AMD has for many years said they're focusing on servers and mobile, since those markets are the most profitable for them and they don't really have the luxury of dabbling in anything else.

Even so, it should be a better match for desktop loads than it is. The trouble appears to be as much in the execution as the philosophy: many investigators have noted that the Bulldozer cache system in particular is higher-latency than Intel's equivalents, among other problems. Had AMD released a stronger implementation, we might be less focused on the shortcomings of its design.

However, it's possible they'll eventually get back to what you're looking for whenever a process shrink gives them more transistors to spend on FP resources -- though they might be shared with the IGP, as ish718 suggests and as AMD's original "Fusion" vision promised long ago.
 
Flatland_Spider
Graphmaster Gerbil
Posts: 1324
Joined: Mon Sep 13, 2004 8:33 pm

Re: Not really an 8-Core CPU then?

Fri Apr 27, 2012 9:07 am

rcs2k4 wrote:
I looked at the CPU die diagrams on how it's all held together - In theory it looks to be a better design choice than HT on Intel's CPU's. We all know in reality thats not the case, but there's definatly 8 core's (of sorts) that run at a nominal speed and will carry out a thread each for 8 in one go.


Hyperthreading and AMD's modules are solutions to two different problems. Hyperthreading is used to cover up stalls in the processor pipeline. It's like when you are doing two tasks at once, but the second task is for when you're waiting on the first.

AMD's module system is about massive amounts of cores. It's a good idea, but there are just technical details that hold it back. A fully loaded module effectively cuts the clockspeed in half because the decode unit alternates in handing out work to the cores. When people say, Bulldozer needs high clockspeeds to function well, that is the reason why.
 
Airmantharp
Emperor Gerbilius I
Posts: 6192
Joined: Fri Oct 15, 2004 10:41 pm

Re: Not really an 8-Core CPU then?

Fri Apr 27, 2012 2:00 pm

I think UberGerbil got it spot on;

Bulldozer is a server-focused architecture being sold to desktop consumers. It's design is also focused on higher-frequency operation, similar to Netburst (but not as crippled), and as noted, improving the cache latencies may go a long way towards increasing the IPC.

To be honest, I did think and still think that Bulldozer's architecture is interesting, and I think that it could be very competitive in the desktop/workstation space with a little improvement.
 
Buub
Maximum Gerbil
Posts: 4969
Joined: Sat Nov 09, 2002 11:59 pm
Location: Seattle, WA
Contact:

Re: Not really an 8-Core CPU then?

Fri Apr 27, 2012 2:04 pm

Flatland_Spider wrote:
AMD's module system is about massive amounts of cores. It's a good idea, but there are just technical details that hold it back. A fully loaded module effectively cuts the clockspeed in half because the decode unit alternates in handing out work to the cores. When people say, Bulldozer needs high clockspeeds to function well, that is the reason why.

I think saying it cuts clockspeed in half is overstating it. AMD chose this design because the decode/dispatch unit isn't constantly 100% busy feeding a single core. It can decode/dispatch faster than the core can execute in most cases, meaning that feeding multiple cores uses the decode/dispatch logic much more efficiently (similar to hyperthreading, ironically, because the rest of the design, as noted above, is not really similar to hyperthreading).
 
ronch
Graphmaster Gerbil
Posts: 1142
Joined: Mon Apr 06, 2009 7:55 am

Re: Not really an 8-Core CPU then?

Sun Apr 29, 2012 1:57 pm

Imagine for a minute that you have 2 K10 cores. Take out each core's fetch and decode units plus their respective L1 instruction caches. In their place, put a beefed up fetch/decode unit and put back the 64KB of L1 I-cache which now services both cores. Let each core able to access its own FPU and the other core's FPU to form a 256-bit unit.

Now stand back. Is it still 2 cores? Or is it a single core with better hyperthreading? I think it's the former. There's no hyperthreading magic here at all. AMD took two cores, compromised them by melding them together to share resources, and called it better than Hyperthreading. Maybe it is, but not for the majority of desktop users.
Last edited by ronch on Mon Apr 30, 2012 2:03 am, edited 1 time in total.
NEC V20 > AMD Am386DX-40 > AMD Am486DX2-66 > Intel Pentium-200 > Cyrix 6x86MX-PR233 > AMD K6-2/450 > AMD Athlon 800 > Intel Pentium 4 2.8C > AMD Athlon 64 X2 4800 > AMD Phenom II X3 720 > AMD FX-8350 > RYZEN?
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Not really an 8-Core CPU then?

Sun Apr 29, 2012 3:40 pm

Not in a strict sense. But I'd say it lies on a continuum between hyperthreading and fully independent cores. Intel's hyperthreading is like a mutant beast with one body and two heads; AMD's modules are like Siamese twins that share a couple of vital internal organs.
Nostalgia isn't what it used to be.
 
ronch
Graphmaster Gerbil
Posts: 1142
Joined: Mon Apr 06, 2009 7:55 am

Re: Not really an 8-Core CPU then?

Mon Apr 30, 2012 1:32 am

There are two cores in a Bulldozer 'module', alright. Two COMPROMISED cores!

Hyperthreading looks like a tech revolution compared to what AMD did with FX. HT is smart enough (and uses far less silicon real estate!) to know which execution resources can be freed up for another thread. It's called finesse. It's called efficiency. With FX, two weak, hot cores are melded together, compromised, then marketed as though it's a technological revolution (CMT, I thnk), which it isn't. If it did reach 6.0GHz at 32nm without sucking more than 125w, THEN it's a revolution.

And while we're on the subject of 'modules', which AMD would have you believe is another technological revolution (remember M-Space?), I'd like to talk a little more about it. What is a module? Is it something that lets you synthesize a core just as easily as 1-2-3? Or, is it something that lets you synthesize a whole processor die in 5 seconds? The way I see it, the FX module hasn't received any M-spacing magic since introduction, so I think it's the latter. What this means, is that you can take a Bulldozer module and synthesize your own little project quite easily. Yeah, easier than baking Apple Pie. But Intel can do that sort of thing too. They can get a Sandy Bridge core, use Paint to orient the cores whichever way they want, add cache and the uncore parts, and voila!! Sandy Bridge-E!!! OMG!!! And over at VIA, the small but hardworking Centaur design team can take a couple of Nano cores, put them beside each other, and presto! Nano X2! Take 2+2 = Via Quad Core!!! Yes, fellas, this is the Platinum Age of computing. And you heard it from me first.

So yes, the FX-8150 is, perhaps controversially, an 8-core processor. It has EIGHT COMPROMISED cores sucking 125w and sold for a pretty penny because AMD is selling you revolutionary 'M-Space' and 'CMT' technologies in a cool, tin box! And did I mention, it has EIGHT cores?

Honestly though, if the FX-8150 is sold for $150 today I just might bite.
NEC V20 > AMD Am386DX-40 > AMD Am486DX2-66 > Intel Pentium-200 > Cyrix 6x86MX-PR233 > AMD K6-2/450 > AMD Athlon 800 > Intel Pentium 4 2.8C > AMD Athlon 64 X2 4800 > AMD Phenom II X3 720 > AMD FX-8350 > RYZEN?
 
geekl33tgamer
Graphmaster Gerbil
Topic Author
Posts: 1023
Joined: Tue Aug 25, 2009 7:25 pm
Location: England
Contact:

Re: Not really an 8-Core CPU then?

Tue May 08, 2012 5:49 pm

ronch wrote:
Honestly though, if the FX-8150 is sold for $150 today I just might bite.

If your that enthusiastic about it to write that humorous reply, you can have one of mine and i'll just go and buy an Ivy Bridge... :wink:
Mega Beast - Intel i7 4790K | Gigabyte Z97X-UD3H-BK | 32GB DDR3 | SLI GTX 1070 | Samsung 850 Evo 1TB
Mini Beast - Intel C2E QX9770 | Gigabyte X48T-DQ6 | 16GB DDR3 | KFA2 LP GTX 750Ti | Seagate 2TB SSHD
 
td1353l
Gerbil
Posts: 28
Joined: Mon May 14, 2012 9:16 pm

Re: Not really an 8-Core CPU then?

Sat May 19, 2012 11:36 pm

nope, kindof like intel, it uses hyperthreading, except its not any faster than last gen, so not much of a upgrade speed wise, minus the fact you now have "8" cores to play with.

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On