AMD CPU speculation... and expert conjecture

[de5_Roy]

New leaked opteron roadmap:

http://tweakers.net/nieuws/93422/gerucht-amd-richt-zich-minder-op-high-end-chips.html

been posted already.
legitimacy aside, the new bits are that amd is pushing berlin cpu (bga) to microservers, populating that segment with 4 different uarches - piledriver, jaguar, steamroller and arm a64/v8/watever. even the ones aimed at compute/media clusters will be bga... so no upgrades...? i am newb about servers, so this move doesn't make much sense to me. i understand why kyoto, berlin and seattle (berlin and toronto cpu) will be bga, being small sized system-on-a-board (is that what it's called?) with own memory and stuff.

another thing caught my eye: HT is specifically mentioned with warsaw but pcie 3.0 is with the rest. what's up with that?

 

[juggernautxtr]

New leaked opteron roadmap:

http://tweakers.net/nieuws/93422/gerucht-amd-richt-zich-minder-op-high-end-chips.html

Thats one thing i never understood why amd doesn't give us this G34 style socket for their cpus, the zif socket is 20+ years old, and it has known connectivity problems. I would be happy to pay a little more for the better socket, they could still keep the backwards compatibility,for those that can't except change.
I still think the zif socket is part of AMD's hold on performance.

 

[jdwii]

I was wondering something: Why can't intel and AMD share the same socket like AMD and Nvidia share the same PCI-E bus? I can't find any reason. Or is it just because it has always been that way?

It would have been a lot easier if the sockets were just like ddr and pci-e.. although the manufacturers would probably sell substantial less motherboards. But it would benefit AMD and Intel at the same time.. you don't need a new motherboard anymore for upgrading a new cpu so the treshold is lower.

Well, Intel uses LGA which is quite different from AMD... can`t they come to some agreement? technically yes... but politically... you wont be seeing that anytime soon.

20 years ago they did share the same socket

 

[juanrga]

That server roadmap was posted and discussed before. It was also explained why AMD is migrating from HTT to PCIe.

This thread is going in circles once again.

 

[juanrga]

I was wondering something: Why can't intel and AMD share the same socket like AMD and Nvidia share the same PCI-E bus? I can't find any reason. Or is it just because it has always been that way?

It would have been a lot easier if the sockets were just like ddr and pci-e.. although the manufacturers would probably sell substantial less motherboards. But it would benefit AMD and Intel at the same time.. you don't need a new motherboard anymore for upgrading a new cpu so the treshold is lower.

The reason why AMD and Intel don't use the same socket for CPUs is the same reason why AMD and Nvidia don't use the same 'socket' for its GPUs and it is the same reason why AMD has different sockets for its different processors. You cannot use an AMD Opteron 6000 series in a FM1 socket, therefore requiring an AMD socket to support Intel chips is technically unreliable. The technologies are completely different.

The only possibility to offer a common CPU socket would be to cut-down the technology up to obtain a minimum common denominator to all processors. This would be useless, expensive, a giant step backwards, and it would also stop evolution of processors.

The same happen with DDR and PCIe but at less scale. E.g. the forthcoming DDR4 standard is pin incompatible with DDR3 because of the evolution of the technology. Therefore you cannot use DDR4 modules in a current motherboard.

 

[FALC0N]

The answer to the socket question was indirectly discussed a few pages back in this very thread.

One of the outcomes that came out of all that legal wrangling over AMD's x86 license was that AMD was granted a perpetual x86 license, but was NOT allowed to copy intel designs. Prior to that, AMD could make a straight copy or something similar enough that it would run on the intel motherboard. That changed with the 1995 settlement. So it's more legal than technical.

 

[FALC0N]

The reason why AMD and Intel don't use the same socket for CPUs is the same reason why AMD and Nvidia don't use the same 'socket' for its GPUs .....

Well, they do both plug into the PCI express slot. So I'm not sure what you mean when you say they don't use the same socket.

The only possibility to offer a common CPU socket would be to cut-down the technology up to obtain a minimum common denominator to all processors. This would be useless, expensive, a giant step backwards, and it would also stop evolution of processors.

The same happen with DDR and PCIe but at less scale. E.g. the forthcoming DDR4 standard is pin incompatible with DDR3 because of the evolution of the technology. Therefore you cannot use DDR4 modules in a current motherboard.

Partially true.

We saw AMD do "Socket A" for a very long time and did it quite well. What ended that was the on-die memory controller. The difference between socket 939 and AM3+ is quite subtle. Mostly power to the cpu, memory controller, and a handful of evolutionary changes. If not for the memory controller, you could have made those parts backwards compatible for the most part. In fact, they did just that with socket AM2 through AM3.

But eventually you run into something like the APU that really does require the platform to migrate to something a significantly different.

 

[jed]

Every business has a budget including the one I own, I had two computer's to update
the first was the E6600 which was replaced by the 2600k and the Q6600 replaced by
the 3770k.
After buying the 2600k I waited for AMD's 8150 and it turned out be nothing like they
said it was supposed to be, so as soon as Intel released the 3770k I purchased it the
next day, no one knew how the 8350 would perform, you know after 8150 hype fest.

Yes, I agree with you. Waiting for the 8350 after the 8150 over-hype was a very hard decision, close to being a fanboi (you can mix hope and some other non-fanboi based arguments ). But, for instance, instead of getting the 3770K, you could have gotten another 2600K on discount on a Z77 or cheap Z68 (I got a 2700k for $150 the day PD came to the stores; can't beat that, haha).

--------------

I had a thought today, a weird one, but what if AMD could go QuadFather all over again with APUs. Slapping an improved HT "ring" to a dual socket for 2xAPU love would be amazing IMO. I would actually think, on one hand, it sounds amazing for bragging rights and doable, since they are going to make servers with that config, right? RIGHT?

A direct interconnect for the APUs would be better, but helluva more expensive per platform and not re-usable. Plus an APU re-design would be needed for it I guess.

The more I think about it, the less plausible it sounds 8(

Cheers!

Thx for being open minded!
Spending 300.00 on a CPU has always been my sweet spot from the day's of my Intel P4 3.2,
I was also thinking of purchasing the 2700k but at the time I wanted something a little different
then the 2600k I already owned, so for me that only left the 3770k, each cpu I purchased had the
best performance in it's class at the time of purchase.
Not for nothing the 2600k is a 3 year old chip that aged pretty damn well if I do say so myself,
it's still competitive with any and every 8 core or 8 thread cpu to this day.

 

[juanrga]

The reason why AMD and Intel don't use the same socket for CPUs is the same reason why AMD and Nvidia don't use the same 'socket' for its GPUs .....

Well, they do both plug into the PCI express slot. So I'm not sure what you mean when you say they don't use the same socket.

Don't confound the 'socket' where the GPU is soldered on the graphics card PCB

with the PCIe bus of the card

You cannot cut a Hawaii GPU from a R9-290 and use it in a 780 Ti, by the same reason that you cannot use an i7-4770k in a FM2 mobo.

 

[jed]

I do have a comparison of i5-3570 (at stock reportedly) to 8350 (OC'd):

http://www.headline-benchmark.com/results/826c8d55-650e-46b7-bd51-8a1e2869dae8/9a411141-526d-4452-8e78-1c9891c2efa0

Now let's be fair and overclock the i5-2500 or i5-3570 sense price is no longer the issue.

http://www.bit-tech.net/hardware/2012/11/06/amd-fx-8350-review/4

http://www.bit-tech.net/hardware/2012/11/06/amd-fx-8350-review/5

 

[Ags1]

Jed, I find those bit-tech results a bit hard to believe. It's all very well to say they did a 'real-world' bench using Gimp, but which parts of Gimp? Seeing how the i3 places relative to the i5 I would say the test is single threaded. They also mention the test results rely on memory and storage performance, which means they are not doing a pure test of the CPU. Which is why I prefer synthetic benchmarks, and of course wrote one myself.

The handbrake result also looked a bit off to me, and I recall Tom's also used handbrake in their FX8350 review, and got totally different results:

http://www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html

 

[gamerk316]

Jed, I find those bit-tech results a bit hard to believe. It's all very well to say they did a 'real-world' bench using Gimp, but which parts of Gimp? Seeing how the i3 places relative to the i5 I would say the test is single threaded. They also mention the test results rely on memory and storage performance, which means they are not doing a pure test of the CPU. Which is why I prefer synthetic benchmarks, and of course wrote one myself.

The handbrake result also looked a bit off to me, and I recall Tom's also used handbrake in their FX8350 review, and got totally different results:

http://www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html

To be fair, real world tasks are hardly ever going to be synthetic.

Also, next time I run the benchmark, I'm going to pay a LOT more attention to the CPU test on where threads are getting assigned. Looking at the 2600k versus 8350 results, I would *expect* the 2600k to win up to 4 threads, start falling behind in 5-8 threads, then pull ahead again above 8 threads. I don't see that.

http://www.headline-benchmark.com/results/c309c1f4-6f00-4def-8127-b8d679fe5989/9a411141-526d-4452-8e78-1c9891c2efa0

Kinda makes me wonder if the HTT cores are getting loaded before the full cores...In any case, for both CPU's, I'd expect close to linear performance gains up to 4 cores [no *major* internal CPU bottlenecks], then lesser gains due to the performance hits of HTT/CMT (with larger gains for AMD from this point forward, eventually passing Intel in performance). Basically, I'm going to check to make sure the cores are loaded on my 2600k in this order: 0,2,4,6,1,3,5,7, loading the HTT cores last during the threading CPU test.

Another thing I might do is force the application affinity to just one core, to see how the 2600k does on each test when threads are fighting over CPU resources. Might be a good *core for core* test to throw in: See how a single core does when 8+ threads are trying to use it.

 

[jed]

Jed, I find those bit-tech results a bit hard to believe. It's all very well to say they did a 'real-world' bench using Gimp, but which parts of Gimp? Seeing how the i3 places relative to the i5 I would say the test is single threaded. They also mention the test results rely on memory and storage performance, which means they are not doing a pure test of the CPU. Which is why I prefer synthetic benchmarks, and of course wrote one myself.

The handbrake result also looked a bit off to me, and I recall Tom's also used handbrake in their FX8350 review, and got totally different results:

http://www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html

I notice every thing you post has Intel at a clock disadvantage of about 600MHz or more,
I used the 2500k and 3570k because they cost about the same as the 8350, with just 4
cores and no HT, now all three cpu's top out at 4.8 - 5.0GHz and here on this forum most
of us overclock everything we get, so with that being said no more of the stock clocks for
the Intel cpu, get some benchmark with both overclocked so we can inform people on just
how good these chips perform against one another.

 

[jdwii]

now all three cpu's top out at 4.8 - 5.0GHz
No way
8350fx 4.6-4.8 some around 5.1(water or at least should be water)
I7 3770K and If 3570K 4.4Ghz
Air cooling

 

[Master-flaw]

Jed, I find those bit-tech results a bit hard to believe. It's all very well to say they did a 'real-world' bench using Gimp, but which parts of Gimp? Seeing how the i3 places relative to the i5 I would say the test is single threaded. They also mention the test results rely on memory and storage performance, which means they are not doing a pure test of the CPU. Which is why I prefer synthetic benchmarks, and of course wrote one myself.

The handbrake result also looked a bit off to me, and I recall Tom's also used handbrake in their FX8350 review, and got totally different results:

http://www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html

I notice every thing you post has Intel at a clock disadvantage of about 600MHz or more,
I used the 2500k and 3570k because they cost about the same as the 8350, with just 4
cores and no HT, now all three cpu's top out at 4.8 - 5.0GHz and here on this forum most
of us overclock everything we get, so with that being said no more of the stock clocks for
the Intel cpu, get some benchmark with both overclocked so we can inform people on just
how good these chips perform against one another.

There are just to many variables...Yea the 3570K is only a bit more than an 8350 but an entry ATX MoBo with O.C.ing is more expensive...I know this as I built the base of my rig on a very limited budget.
Z77 board with a 3570K was around 330-350...the 970 board I got(not the best but has served me well) and an 8350 was only 250(260 without MiR)...the price difference is actually a lot bigger than people give it credit for.
Also cooling comes into play...when comparing these chips O.C.ed I think the price of cooling should be considered. AMD can do 4.8 on air.

 

[juggernautxtr]

Jed, I find those bit-tech results a bit hard to believe. It's all very well to say they did a 'real-world' bench using Gimp, but which parts of Gimp? Seeing how the i3 places relative to the i5 I would say the test is single threaded. They also mention the test results rely on memory and storage performance, which means they are not doing a pure test of the CPU. Which is why I prefer synthetic benchmarks, and of course wrote one myself.

The handbrake result also looked a bit off to me, and I recall Tom's also used handbrake in their FX8350 review, and got totally different results:

http://www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html

To be fair, real world tasks are hardly ever going to be synthetic.

Also, next time I run the benchmark, I'm going to pay a LOT more attention to the CPU test on where threads are getting assigned. Looking at the 2600k versus 8350 results, I would *expect* the 2600k to win up to 4 threads, start falling behind in 5-8 threads, then pull ahead again above 8 threads. I don't see that.

http://www.headline-benchmark.com/results/c309c1f4-6f00-4def-8127-b8d679fe5989/9a411141-526d-4452-8e78-1c9891c2efa0

Kinda makes me wonder if the HTT cores are getting loaded before the full cores...In any case, for both CPU's, I'd expect close to linear performance gains up to 4 cores [no *major* internal CPU bottlenecks], then lesser gains due to the performance hits of HTT/CMT (with larger gains for AMD from this point forward, eventually passing Intel in performance). Basically, I'm going to check to make sure the cores are loaded on my 2600k in this order: 0,2,4,6,1,3,5,7, loading the HTT cores last during the threading CPU test.

Another thing I might do is force the application affinity to just one core, to see how the 2600k does on each test when threads are fighting over CPU resources. Might be a good *core for core* test to throw in: See how a single core does when 8+ threads are trying to use it.

Jed, I find those bit-tech results a bit hard to believe. It's all very well to say they did a 'real-world' bench using Gimp, but which parts of Gimp? Seeing how the i3 places relative to the i5 I would say the test is single threaded. They also mention the test results rely on memory and storage performance, which means they are not doing a pure test of the CPU. Which is why I prefer synthetic benchmarks, and of course wrote one myself.

The handbrake result also looked a bit off to me, and I recall Tom's also used handbrake in their FX8350 review, and got totally different results:

http://www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html

I notice every thing you post has Intel at a clock disadvantage of about 600MHz or more,
I used the 2500k and 3570k because they cost about the same as the 8350, with just 4
cores and no HT, now all three cpu's top out at 4.8 - 5.0GHz and here on this forum most
of us overclock everything we get, so with that being said no more of the stock clocks for
the Intel cpu, get some benchmark with both overclocked so we can inform people on just
how good these chips perform against one another.

I agree, even in a multifaceted (different test at the same time) program wouldn't the operating system recognize this and schedule accordingly.
I am no programmer, but I see the single program running with several tasks is still only one program with one compiler.
real world would be 3-4 different programs all running their own compiler and demanding cpu time.

maybe you can clarify me,I could be wrong in my understanding of programming,but the single program as far as i know can't run more than one compiler,yes different instructions sets called for, but its all done in order of call? where as the separate programs would all be demanding their instruction sets at the same time.

 

[de5_Roy]

There are just to many variables...Yea the 3570K is only a bit more than an 8350 but an entry ATX MoBo with O.C.ing is more expensive...I know this as I built the base of my rig on a very limited budget.
Z77 board with a 3570K was around 330-350...the 970 board I got(not the best but has served me well) and an 8350 was only 250(260 without MiR)...the price difference is actually a lot bigger than people give it credit for.
Also cooling comes into play...when comparing these chips O.C.ed I think the price of cooling should be considered. AMD can do 4.8 on air.

tht's a pretty good deal for the fx combo. i see the same parts costing $280-290 sans mir.

 

[Master-flaw]

There are just to many variables...Yea the 3570K is only a bit more than an 8350 but an entry ATX MoBo with O.C.ing is more expensive...I know this as I built the base of my rig on a very limited budget.
Z77 board with a 3570K was around 330-350...the 970 board I got(not the best but has served me well) and an 8350 was only 250(260 without MiR)...the price difference is actually a lot bigger than people give it credit for.
Also cooling comes into play...when comparing these chips O.C.ed I think the price of cooling should be considered. AMD can do 4.8 on air.

tht's a pretty good deal for the fx combo. i see the same parts costing $280-290 sans mir.

The thing is with the 8350 is the thing retails at $200 but it goes on sale more often then not...
I waited 2 months for a better deal on the 3570K and the lowest it got was $200(was gunna pull the trigger at $180 with a cheap 7870)...It NEVER goes on sale...that things price is as rigid as it gets. I got my 8350 on sale for $180 after it dropped a few times...it had a 10 MiR with it along with the Asus m5a97 le 2.0 MoBo I was a bit nervous to go with for $80(other users told me it was fine)...I couldn't argue with it...
The I5's are faster, but when it comes down to it, they are quite a bit more expensive when building...

 

[jed]

now all three cpu's top out at 4.8 - 5.0GHz
No way
8350fx 4.6-4.8 some around 5.1(water or at least should be water)
I7 3770K and If 3570K 4.4Ghz
Air cooling

First of all the i7 3770k wasn't mentioned in the above paragraph not once,
in my experience the 2500k would overclock a little better then the 2600k I
owned, which by the way ran @ 4.7 with air cooling and @ 5.2 with water.
The three cpu's I used was the 2500k, 3570k and 8350 with all of them
using aftermarket cooling, weather it be air or water cooling.

 

[jed]

There are just to many variables...Yea the 3570K is only a bit more than an 8350 but an entry ATX MoBo with O.C.ing is more expensive...I know this as I built the base of my rig on a very limited budget.
Z77 board with a 3570K was around 330-350...the 970 board I got(not the best but has served me well) and an 8350 was only 250(260 without MiR)...the price difference is actually a lot bigger than people give it credit for.
Also cooling comes into play...when comparing these chips O.C.ed I think the price of cooling should be considered. AMD can do 4.8 on air.

tht's a pretty good deal for the fx combo. i see the same parts costing $280-290 sans mir.

The thing is with the 8350 is the thing retails at $200 but it goes on sale more often then not...
I waited 2 months for a better deal on the 3570K and the lowest it got was $200(was gunna pull the trigger at $180 with a cheap 7870)...It NEVER goes on sale...that things price is as rigid as it gets. I got my 8350 on sale for $180 after it dropped a few times...it had a 10 MiR with it along with the Asus m5a97 le 2.0 MoBo I was a bit nervous to go with for $80(other users told me it was fine)...I couldn't argue with it...
The I5's are faster, but when it comes down to it, they are quite a bit more expensive when building...

Right now you can get them both for a decent price, if your close to a microcenter,
a 10.00 difference in price.

http://cart.microcenter.com/cart.aspx?RedirectUrl=http://www.microcenter.com

 

[Ags1]

Jed, I find those bit-tech results a bit hard to believe. It's all very well to say they did a 'real-world' bench using Gimp, but which parts of Gimp? Seeing how the i3 places relative to the i5 I would say the test is single threaded. They also mention the test results rely on memory and storage performance, which means they are not doing a pure test of the CPU. Which is why I prefer synthetic benchmarks, and of course wrote one myself.

The handbrake result also looked a bit off to me, and I recall Tom's also used handbrake in their FX8350 review, and got totally different results:

http://www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html

I notice every thing you post has Intel at a clock disadvantage of about 600MHz or more,
I used the 2500k and 3570k because they cost about the same as the 8350, with just 4
cores and no HT, now all three cpu's top out at 4.8 - 5.0GHz and here on this forum most
of us overclock everything we get, so with that being said no more of the stock clocks for
the Intel cpu, get some benchmark with both overclocked so we can inform people on just
how good these chips perform against one another.

Agreed, but currently all my FX results are OC'd and most of the Intel ones aren't. Looks like a different kind of user buys an FX and they are highly likely to OC. I will post an OC'd 3570 if I ever get one, or a stock 8350, if I ever get one.

 

[blackkstar]

now all three cpu's top out at 4.8 - 5.0GHz
No way
8350fx 4.6-4.8 some around 5.1(water or at least should be water)
I7 3770K and If 3570K 4.4Ghz
Air cooling

First of all the i7 3770k wasn't mentioned in the above paragraph not once,
in my experience the 2500k would overclock a little better then the 2600k I
owned, which by the way ran @ 4.7 with air cooling and @ 5.2 with water.
The three cpu's I used was the 2500k, 3570k and 8350 with all of them
using aftermarket cooling, weather it be air or water cooling.

I'm not saying that Haswell and IB can't hit higher clocks, however...

Intels' CPUs have been gradually growing less likely to hit higher clocks. You can find Haswells close to 5ghz if you dig across the internet, and then compare that to SB and then say that they overclock the same.

But if you have 1000 Sandy Bridges, 1000 Ivy Bridges, and 1000 Haswells, you're going to have a lot more Haswells that don't overclock very well compared to how many good clocking Sandy Bridges you'll see.

That is the point we're getting at. I don't think I've ever seen a Piledriver chip that wouldn't go past 4.4ghz no matter how much voltage you threw at it (temps are a different matter, but temps can be fixed while there's nothing you can do about a chip that just wants vcore). You do find Haswells that won't go 4.4ghz and you even find IBs that don't OC very well at all.

Unless I'm mistaken, you're pretty much guaranteed 4.7ghz with a piledriver chip if you have good enough cooling, and if you're on water you're pretty much guaranteed 5ghz. The golden FX 8350s are the ones that hjit 5ghz+ on air, and the golden IBs and Haswells hit high 4ghz range.

I'm quite sure you can find people running Haswell and IB at higher vcore claiming to go fast, but the chips don't last. I knew an Intel fanboy on OCN who had a 5.2ghz delidded SB he ran 1.55v through. He boasted about it for a year and called it the Vishera killer, and then one day he put the whole thing up for sale, parted out, and then downgraded from what he had. In other words, he probably degraded the chip and went to unload it on some poor, unsuspecting OCN member.

 

[griptwister]

I really like what I'm seeing from Kaveri... I already have a FX 8350 and I would like to maybe build a second rig with Steamroller CPU. I have enough money to purchase one of these bad boys... almost enough for a Kav APU too. I think I might wait a bit though for the Actual Kaveri CPUs to come out before I buy anything. The direct upgrade path to Excavator sounds promising too.

http://www.newegg.com/Product/Product.aspx?Item=N82E16813128655&Tpk=FM2%2b%20UP4

 

[gamerk316]

now all three cpu's top out at 4.8 - 5.0GHz
No way
8350fx 4.6-4.8 some around 5.1(water or at least should be water)
I7 3770K and If 3570K 4.4Ghz
Air cooling

First of all the i7 3770k wasn't mentioned in the above paragraph not once,
in my experience the 2500k would overclock a little better then the 2600k I
owned, which by the way ran @ 4.7 with air cooling and @ 5.2 with water.
The three cpu's I used was the 2500k, 3570k and 8350 with all of them
using aftermarket cooling, weather it be air or water cooling.

2600k can get to around 4.6/4.7 or so, sometimes 4.8. So about the same as FX. IB/Haswell clock slightly lower, but have higher IPC, so they should perform about the same at max OC.

 

[gamerk316]

CPU test of headline has some bugs to work out. HTT is mucking with the CPU results, like I thought it was.

Look at it logically: If I have 4 cores, and I do the same amount of work in 4 threads, you would expect linear absolute performance gains up to 4 threads, correct?

Taking a look at two 2600(k) runs:

http://www.headline-benchmark.com/results/c309c1f4-6f00-4def-8127-b8d679fe5989
http://www.headline-benchmark.com/results/93a871e9-4a76-42aa-8f91-ef08fedef7dc

Notice the various dropoffs in performance as threads increase? That shouldn't be happening, unless multiple threads are getting stuck on the same core, OR a HTT core is getting used before a "full" core, and the shared frontend is causing massive performance drops.

Compare that to this run:

http://www.headline-benchmark.com/results/359025ad-a71c-4df1-a21e-2dcac265a956/87080cac-179d-4a2c-a2a9-cb1e90667a3f

Where I disabled the HTT cores, and compared against my last benchmarks. Note how the performance dropoffs vanished, and in some cases, performance improved by DISABLING cores?

Now, we still have an issue where performance stalls. What I *think* is going on, is multiple threads are going to the same core, so performance flatlines, then jumps by double the next time a thread gets assigned, maxing at about 5 threads (remember: HTT cores disabled). So in the case of the Integer test, I think the treads are being assigned like this:

Thread 1: Core 0
Thread 2: Core 1
Thread 3: Core 0/1
Thread 4: Core 2
Thread 5: Core 3

Hence why performance stalls on three threads, then increases linearly up to 5 (which doesn't make sense for a quad). Same thing likely happening in the FP test:

Thread 1: Core 0
Thread 2: Core 0
Thread 3: Core 1
Thread 4: Core 1
Thread 5: Core 2

Based on the huge jump going from 4 to 5, I'd suspect one of the other threads gets reassigned to core 3 at some point, but I'd need to break out my low level debug tools to prove this.

Ags1, I'd recommend that you manually assign threads to cores for this test. Start with the Even (Non HTT/CMT) cores, then proceed to the Odd ones. Once you have more threads then cores, don't bother anymore (really doesn't matter at that point). That should remove the Windows scheduler from affecting performance...Right now, due to HTT having a much higher performance cost if used wrong, Intel's numbers are probably grossly understated in your benchmark. [Can't wait to see the response to THAT one.]