K8 vs K10 on 790FX - Results

Firstly. Big appologies for the differences in RAM timings between the Windsor and Agena in the following.

Due to problems with this board I could not run CAS4 on the K8 on this bios at least. Also I should have by rights used a 10x multi to keep RAM frequency the same.

Instead though I've clocked the K8 20Mhz faster in an attempt to keep things as even as possible. In the end though the aim of the exersize was find areas were K10 majorly outperforms K8 not nitpick at anything less than 2%. Anyone who's spent time with Phenom on the current BIOS will no what a pain the arse it really is atm to run overclocked. Hence someitmes it's just too hard to get things exactly how you want them. I will keep at it and RERUN THE BENCHMARKS with identical settings.

Now, the focus here is on 1-2 threaded Apps only, with the aim of giving people a better idea whether an X4 is for them at the current clocks and rev's, especially those with current X2s I

THIS IS NOT AN INTEL VS AMD THREAD. Especially given it's single-dual thread nature, where we allready are pefectly aware K10 does not have the IPC of Core for MOST desktop app's (but not all, I will say that) Trolls, please See the troll thread, not this one. That goes for anyone

Also if someone with a good brain can have a check over my % calcs, we all make mistakes.




Apart from the minor RAM timing differences, all tests are on the same machine, same install. All run on the Phenom first. Some repeated at the end to ensure no degredation of the OS had occured during the install of some programs.

Some I sitll have to do for K8 Include Superprime and the new release Wprime. stay tuned for that.

Setup:

Mainboard: Gigabyte MA-790FX-DQ6 Bios F2H
RAM: Crucial Ballistix 1066
Phenom: 4-4-4-12 487Mhz
Athlon: 5-4-4-12 440Mhz
Video: Gecube HD3850 256mb @ 750/1000

CPUs:

K8 Windsor 1Mbx2 L2 @ 2695Mhz
K10 Agena @ 2680Mhz

OS
Windows XP32 SP2 +crit updates


On with the results...


Super Pi 1M


http://img155.imageshack.us/img155/1868/k10pirarlk7.jpg http://img520.imageshack.us/img520/9955/k8pirarpc2.jpg

Phenom: 27.687s
Athlon:30.328s
K10 +9.5%

*Note, Winrar bench not run single threaded- discard. Will rerun along with others

-------------------------------
Nuclearus MC:

http://img47.imageshack.us/img47/568...clearuslm0.jpghttp://img444.imageshack.us/img444/5...clearusjw9.jpg

Phenom:
ALU: 4973
FPU: 5116

Athlon:
ALU: 4356
FPU: 5067

ALU K10 +14%
FPU K10 +0-1%

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

Lightmark 2007

http://img338.imageshack.us/img338/5...ghtmarklk1.jpghttp://img138.imageshack.us/img138/2...ghtmarkog4.jpg

Phenom: 269.3
Athlon: 233.7

K10 +15%

--------------------------------
LAME 3.7 MP3 192Kbs - Standard Compiler

http://img146.imageshack.us/img146/4...lamestdfk1.jpghttp://img527.imageshack.us/img527/1...lamestdfu0.jpg


Phenom: 2:06
Athlon: 2:27

K10 +16%

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

LAME 3.7 MP3 192Kbs - Intel Compiler

http://img155.imageshack.us/img155/8...meintelbc2.jpghttp://img47.imageshack.us/img47/733...meintelnn9.jpg

Phenom: 1:42
Athlon: 2:01

K10 +18%

------------------------------
FLAC CDex 1.70

http://img47.imageshack.us/img47/3879/k10flachm4.jpghttp://img155.imageshack.us/img155/1966/k8flacoo0.jpg

Phenom: 4:17
Athlon: 5:00

K10 +16-17%

-------------------------------
CineBench R10 32bit
http://img146.imageshack.us/img146/4...nebenchpu2.jpghttp://img138.imageshack.us/img138/8...nebenchtm1.jpg

Single Thread:
Phenom:2372
Athlon:2314

K10 + 2.5%

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

3Dmark 06:

http://img516.imageshack.us/img516/2721/76511663zw0.jpghttp://img132.imageshack.us/img132/5...k06cas5rt7.jpg

Phenom:
SM2.0: 4309
SM3.0: 4780

Athlon:
SM2.0: 4142
SM3.0: 4578

SM2 K10 +4%
SM3 K10 +4%


*important to note this Test is not overly CPU bottlenecked. more interest factor


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

Aquamark 3

http://img147.imageshack.us/img147/3853/k10am3oz3.jpghttp://img146.imageshack.us/img146/4531/k8aquaai0.jpg

Phenom:
GFX: 22566
CPU: 13422

Athlon
GFX: 19980
CPU: 12971

GFX: K10 +13%
CPU: K10 +3.4%
Overall score K10 +8%


TO BE CONTINUED..

STILL TO COME

*Video Encoding

*Archving

*SSE heavy Stuff

HOPEFULLY some real games if I can get some help on DEMO's I can use, as I don't have much unfortunatly


*** - UPDATED: 7dec 07 ***


Notes:

RAM timings vary a bit on these ones, I had initially got them both running at CAS4 finally, but then during a CPU swap I've managed to end up with Phenom on Cas 5 for some of these tests, :shakes: :shakes: Frequencies now identical though. So Phenom slightly disadvantaged this time around. Again though we're talking a percent at the most. K10 especially much less affected by the odd timing difference.

=======================

x264 Benchmark

http://img266.imageshack.us/img266/1...2threadft2.jpghttp://img266.imageshack.us/img266/5...2threaddm2.jpg

Phenom:
PASS1:89.00
PASS2:17.58

[106.58]

Athlon:
PASS1:75.13
PASS2:17.97

[93.1]


K10 +14.4% overall

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

Systool

http://img513.imageshack.us/img513/2...systoolpm4.jpghttp://img513.imageshack.us/img513/1...systoolhp2.jpg

Phenom: 263sec
Athlon: 237sec

K10 + 10.9%



notes: CAS4 on K8

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

]Wprime 1.55

http://img461.imageshack.us/img461/4...2threadjv8.jpghttp://img461.imageshack.us/img461/9...2threadrt0.jpg

Phenom: 29.562
Athlon: 32.14

K10 +10.3%

-----------------------------------------
POV-RAY 3.7 single thread

http://img513.imageshack.us/img513/3...371coremk0.jpghttp://img522.imageshack.us/img522/9...371corely8.jpg

Phenom: 394.8
Athlon: 370.5

K10 +6.55%

------------------------------------------
ScienceMark

http://img261.imageshack.us/img261/9...ncemarktv8.jpghttp://img451.imageshack.us/img451/6...encmarkcn4.jpg

Phenom: 1687
Athlon: 1512

K10 +11.6%

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



***Lightman's Results****


Lightman has also kindly let me merge his results from the thread into this post, Enjoy:



I did some tests for you guys simulating Phenom X2 2200MHz with DDR-800 4-4-4:

http://img410.imageshack.us/img410/6...ueenxo3.th.png
CPU Queen 4395 vs 4262 of X2 4200+ 2.2GHz

http://img410.imageshack.us/img410/1...orxxtl8.th.png
CPU PhotoWorxx 26375 vs 17902 of X2 4200+ 2.2GHz

http://img410.imageshack.us/img410/2...zlibgf7.th.png
CPU ZLib 29487KB/s vs 26238KB/s of X2 4200+ 2.2GHz

http://img410.imageshack.us/img410/3...uliars6.th.png
FPU Julia 4660 vs 3742 of X2 4200+ 2.2GHz

http://img135.imageshack.us/img135/2...ndelea5.th.png
FPU Mandel 2308 vs 2114 of X2 4200+ 2.2GHz

http://img135.imageshack.us/img135/3...uliasf7.th.png
FPU SinJulia 5671 vs 5872 of X2 4200+ 2.2GHz

http://img530.imageshack.us/img530/3...erpigl4.th.jpg
Cinebench 2003 x64 + SuperPi




POST #2


http://www.xtremesystems.org/forums/...0&d=1196546523

Good work brother:toast: Nice to see some people posting up some AMD stuff:) Too bad these 6400+ don't seem to O.C much but the bandwidth on the new AMD platform ROCKS!!!!!!!!:up:

http://community.herinteractive.com/...es/popcorn.gif

That's where you've been hiding then...

Good work mAJORD.

BTW I just killed my mobo!

It makes you wonder what AMD was doing that whole time while they were "developing Barcelona"... It seems Intel achieve more in going from Conroe to Penryn than AMD has going from Windsor to Barcelona. This is what AMD brought to the next-gen fight?

Thank you for the work mAJORD. :)

Some details you left out I'd like to now please:

What volts were you running, HT, NB, CPU?
Whats max ref. HT, HT speed and NB speed you reached?

My HT works at half the speed of that and NB is much lower too.

Can you run some Sandra, EVEREST, Systool 16M Pi, ScienceMark and POV-Ray please. :D I'll need to compare later.

Games would be good for those who are interested because I have none here but HL2 and UT as I haven't played since coming here.

My setup at those speeds are giving lower results because I can't run dual channel, high NB or high HT. Bugged out very badly.

Nice post! Look forward to more!

:clap:

Aussie: What the hell? how? :(

KTE, will do.

Volts are as shown in CPU-Z (reporting correct) 1.31 on air.

243 is the max HTT I can reach, it's a hard wall. F3A doesn't help. I hope we can (all) get around this issue as I'd love to put all this cooling to work :( As it is i have no reason to use anything more than the dinky little Arctic64 pro..

Quote:

---

Originally Posted by EnJoY

It makes you wonder what AMD was doing that whole time while they were "developing Barcelona"... It seems Intel achieve more in going from Conroe to Penryn than AMD has going from Windsor to Barcelona. This is what AMD brought to the next-gen fight?

---

I know its quite disappointing isn't it Alex?:(

Quote:

---

Originally Posted by mAJORD

Aussie: What the hell? how? :(

---

PM sent. :D

But C2Q is about 10% faster clock for clock than Phenom and more cheap.

And K10 is about 15% faster clock for clock than K8.

Penryn performs about 15% more clock for clock than K10, and 30% than K8.

Where is the AMD classic price-performance? :(

Quote:

---

Originally Posted by EnJoY

It makes you wonder what AMD was doing that whole time while they were "developing Barcelona"... It seems Intel achieve more in going from Conroe to Penryn than AMD has going from Windsor to Barcelona. This is what AMD brought to the next-gen fight?

---

here we go again.........:rolleyes:

not that difficult to increase that much performance after so many years of netburst crap and from con to pen its just the apps that benefit from additional cache and were already constrained in conroe on fsb. (and the sse4 offcourse)


for the k10 part
k10 first original design was killed because the market was not ready for it.

second k10 design (this one) has some advanced memory control, shared cache and native but for the rest the overestimated the increased performance by that i think. if l3 would be as fast as and as big as intel cache you would see a much better int performance, lets see what shangai brings.

and it was late to the market because they had to delay 65nm because of first soi issues (long time ago, yields ) and native design.

maJORD thx for the info, but aren't you the first to report k8 issues on the 790fx chipset? or did i miss some posts...... i know there were phenom bios issues.

duploxxx

the CAS 4 not sticking is the first problem i've had with it on K8.

Fantastic work mAJORD! I'm jealous of your CPU :) Mine can do 2.5GHz stable :p:

I will throw a bit as well if you happy with that:

http://www.xtremesystems.org/forums/...0&d=1196546523
You can see Clock difference between exactly same score of 508 points in Single threaded test for Phenom and K8 Brisbane :rolleyes:

Quote:

---

Originally Posted by duploxxx

k10 first original design was killed because the market was not ready for it.

---

care to elaborate?

Quote:

---

f l3 would be as fast as and as big as intel cache you would see a much better int performance

---

AMD cpus do not need the extra cashe becuse the have a on-board memory controler. but it would be nice to see it a bit faster. as well as the hole thing with SOI fixed.

agreed, this is ridiculous! Conroe had huge gains over netburst and k8 alike and that's what a new architecture should offer, and from what we've heard, nehalem will be to conroe as conroe was to netburst so another massive gain, but only 1-20% faster for k10 over k8? That's pathetic! Especially with all the delays we experienced.

I don't care what amd does to fix this, but something needs to be done. If bulldozer fails, amd might as well force Hector to stay and focus on how they can lose even more market share and stock value

Netburst->Conroe was a bit different cos Netburst sucked so much that it was easier to get huge performance gains. K8 was much better than Netburst so it's not so easy to get that big performance gain. Though I agree that K10 isn't exactly a succes.

great test majord

Quote:

---

Originally Posted by Nazu

Netburst->Conroe was a bit different cos Netburst sucked so much that it was easier to get huge performance gains. K8 was much better than Netburst so it's not so easy to get that big performance gain. Though I agree that K10 isn't exactly a succes.

---

still, I know an intel engineer and he's been saying for a long time now the increase of conroe-->Nehalem will be similar to that of netburst-->conroe. I know intel has an awesome amount of resources, but still, amd should easily be able to pull 20% across the board over k8 especially with the 4 years it took for them to design k10, and if they wanted that market share and actually be able to call their product an innovation, k10 should have been at least 15+% superior to conroe across the board, and a lot stronger in areas where amd shines like fpu

very good work majord. Finally get to see some interesting result. A 15% gain over the K8 looks about the main in single/double threaded stuff.

Now, if they can get the speeds ramped and keep the prices down it could be a decent cpu on the desktop. It certainly looks like AMD went out of their way to produce a server chip though in my opinion (which at higher speeds seems to be very competitive).

clock for clock its faster, but its also adding 2 more cores. k8 vs k10 that is

maybe try rebenching both at stock speeds?

Quote:

---

Originally Posted by Morais

http://community.herinteractive.com/...es/popcorn.gif

---

Indeed, I'll be sitting back to see the results that follow! *excited*

Nice work mAJORD :up:

Quote:

---

Originally Posted by AliG

still, I know an intel engineer and he's been saying for a long time now the increase of conroe-->Nehalem will be similar to that of netburst-->conroe. I know intel has an awesome amount of resources, but still, amd should easily be able to pull 20% across the board over k8 especially with the 4 years it took for them to design k10, and if they wanted that market share and actually be able to call their product an innovation, k10 should have been at least 15+% superior to conroe across the board, and a lot stronger in areas where amd shines like fpu

---

First,there is almost no chance that that Nehalem will be faster than Conroe as much as Conroe was faster than Netburst,in genereal "average" case.
Nehalem is a Core based uarch. with some (yes radical) changes to system arch as IMC and QuickPath.Conroe/Penryn already have great ALUs and i can see Nehalem some 15% faster there.In FP there could be some greater gains since it will be "2 threads per core" capable design (HyperThreading),but this could be probably seen more in some server applications rather than client(desktop) ones.Intel will lower mem. latency and improve Core uarch overall,but to get anywhere close to 50% faster(in what?SSE?int?) as it was general case in Netburst Vs Core2,there is little chance.Like i said, in some mem. constraint server applications where mem. bandwidth is crucial and where good scaling to more cores is needed,Nehalem could be that much better than Core2.Spec_fp(rate) is a good example where "40% better than Core2" could be achieved.But in this case,we have K10 beating Core2 by as much as 40%.

@mAJORD

Impressive work man :).
What are your observations wrt the NB speed(ie. 2 cases where we have same clock for the cores but with different clocks for NB)?Does L3 latency drop (happens with NB clock increase) help in any measurable way?

Quote:

---

Originally Posted by AliG

still, I know an intel engineer and he's been saying for a long time now the increase of conroe-->Nehalem will be similar to that of netburst-->conroe. I know intel has an awesome amount of resources, but still, amd should easily be able to pull 20% across the board over k8 especially with the 4 years it took for them to design k10, and if they wanted that market share and actually be able to call their product an innovation, k10 should have been at least 15+% superior to conroe across the board, and a lot stronger in areas where amd shines like fpu

---

Actually, Barcelona/Phenom is quite good at floating point. Take a look at this link:

http://www.cisl.ucar.edu/dir/CAS2K7/...torricelli.pdf

A 2 GHz Barcelona is beating a 2.33 GHz Xeon by 30-50% in floating point when using multi-core, heavy memory type engineering applications such as CFD (Fluent) and structural mechanics (LS-DYNA). Take a look at the differences when all four cores are being utilized. SuperPi fits into the cache of Core-2 Duo, hence it will always have better SuperPi scores.

Funny, I have not seen SciMark 2.0 benchmarks in any of the website reviews.

In the type of applications used by servers and engineering analysis, AMD is quite competitive.

The 10% difference in integer performance is nothing. I recall when different steppings of K8 (Venice vs Winchester) gave almost this much improvement. The big advantage of Core 2 Duo is their excellent over-clocking ability. I am sure that AMD will improve in this area, but I am not sure they will ever beat Intel in overclocking.

informal: nehalem is not a core μarch, its a completely different proccess.
Penryn is a core μarch shrinked derivative.
Dont forget tick-tock model..

edit>@trueblue: lsdyna, mathematica, pam crash and fp in general are all memory intensive applications, hence imc has an advantage on that. However this is just a 10% of the pie, amd should have been improved to the 90%.

Quote:

---

Originally Posted by Nazu

Netburst->Conroe was a bit different cos Netburst sucked so much that it was easier to get huge performance gains. K8 was much better than Netburst so it's not so easy to get that big performance gain. Though I agree that K10 isn't exactly a succes.

---

K10 can be a success, but AMD needs to go back to their roots. They need to come up with an impressive pricing scheme. They also need to drop plans for single core chips based on K10. It's completely unnecessary. Not to mention that even dual-core on K10 would be unnecessary at this point. Tri-core and quad-core is what they should offer. The tri-core is something that Intel can't compete with, and dual-core and tri-core would be equally pricey to manufacture and deliver, given that AMD stupidly went for a monolithic design rather than a modular one.

Regardless, K10 seems like a decent option for K8 users. I wouldn't mind getting one so long as the price is right and it works in my board.

There have been a couple of really nice Phenom threads, this being one... kudos.

Quote:

---

Originally Posted by AliG

still, I know an intel engineer and he's been saying for a long time now the increase of conroe-->Nehalem will be similar to that of netburst-->conroe. I know intel has an awesome amount of resources, but still, amd should easily be able to pull 20% across the board over k8 especially with the 4 years it took for them to design k10, and if they wanted that market share and actually be able to call their product an innovation, k10 should have been at least 15+% superior to conroe across the board, and a lot stronger in areas where amd shines like fpu

---

I disagree, looking over mAJORD data, pretty much in agreement with the reviews that did similar, we are looking at 4 to 18% IPC improvement over K8 -- This is healthy, the K10 design rev from that point of view is successful.

If you compared Conroe (Core 2) over Yonah (Core) you get the same kind of 4 to 20% ish type of improvements depending on the app... any design that achieves ~ 10-15% (on average) improvement IPC wise is a good. What made Conroe shine so much was not the improvement over the prior tech along it's same lineage, but the improvement over the segment product which was along a different lineage... then again, it wasn't too difficult to improve over Prescott.

Quote:

---

Originally Posted by fotme

informal: nehalem is not a core μarch, its a completely different proccess.
Penryn is a core μarch shrinked derivative.
Dont forget tick-tock model..

edit>@trueblue: lsdyna, mathematica, pam crash and fp in general are all memory intensive applications, hence imc has an advantage on that. However this is just a 10% of the pie, amd should have been improved to the 90%.

---

You are wrong..It IS based on 4-issue Core2,but it will have improved core(especially SSE part;Hyperthreading included).The most radical change will be system arch. which will change from FSB based to IMC+QuickPath(pretty much like IMC+DirectConnect in K8&K10).So,yes,it is a "tock" but it's not radically new "core-wise",more radical "system arch.-wise".
Here you go:
http://aycu07.webshots.com/image/373...5101578_rs.jpg

Now can we stop this off topic discussion and pay more attention at the subject of this topic?
To me it seems that Phenom in this present form is very good (solid) improvement over K8 in ALU perf.(around 15%).In pure x87 code,there is no improvement(no surprise there),while in SSE intensive cases we can see some very good improvements.The things that bother Phenom the most at present are some bugs(will get fixed soon) and CnQ2 not properly running in some cases.I reckon that with B3 we could finally see some higher clocked nortbridges in Phenoms,since 2Ghz is fairly low and affects the L3 latency in a bad way.

If we look at the K10 as just a tweaked K8, the gains are not bad. But when compared to the competing Conroe and upcoming 45nm (don't know the code names), things doesn't look good.

Quote:

---

Originally Posted by informal

I reckon that with B3 we could finally see some higher clocked nortbridges in Phenoms,since 2Ghz is fairly low and affects the L3 latency in a bad way.

---

Yes i hope for higher clocks too. What do you think about the bugs, will rev b3 be enough? I think we would need at least b4 to get the phenom really "shine"

Quote:

---

Originally Posted by awdrifter

If we look at the K10 as just a tweaked K8, the gains are not bad. But when compared to the competing Conroe and upcoming 45nm (don't know the code names), things doesn't look good.

---

Just a tweaked K8? hahah I had to laugh. Do you really believe it took like 2-3 years to tweak something based on a same core? I doubt it. K10 will grow in time just be patient and you will see how great the application will respond to it that support multi-cpu.

Thanks for sharing mAJORD.

I agree with some others here, makes you wonder what AMD has been doing all this time, developing K10 or having a coffée break? :D

Not that I thought there would be much to improve from K8, but a say a ~15% clock for clock performance boost sure is a bit dissapointing as it means it's a bit behind C2D. AMD just needs to drop this old milked cow and start from scratch. At least consumers can look forward to cheap quads this time around. :D

Quote:

---

Originally Posted by RedBull78

Just a tweaked K8? hahah I had to laugh. Do you really believe it took like 2-3 years to tweak something based on a same core? I doubt it. K10 will grow in time just be patient and you will see how great the application will respond to it that support multi-cpu.

---

Dude, just look at the results. Single threaded performance is barely better than that of K8. The only reason performance gains of k10 over K8 are even in the double digits are due to having double the cores and more cache.

Quote:

---

Originally Posted by EnJoY

Dude, just look at the results. Single threaded performance is barely better than that of K8. The only reason performance gains of k10 over K8 are even in the double digits are due to having double the cores and more cache.

---

ALU scores are actually ~15% better than on K8(per core,not multi core).This is really a good achievement,since Core2 over Core was also 15-20%.
SSE units and cache widths in K10 are doubled when compared to K8 and here you'll see a lot more than15% gain(provided you run vector SSE optimized application-Cinebench10 uses scalar SEE for instance,so not much gained there).
Like Jack said,15% better IPC-wise than K8 is a really good result!Or else you can call Core2 over Core a bad IPC improvement too..

Quote:

---

Originally Posted by EnJoY

Dude, just look at the results. Single threaded performance is barely better than that of K8. The only reason performance gains of k10 over K8 are even in the double digits are due to having double the cores and more cache.

---

Care to post the results you're looking at. :rofl:

Give EnJoY a break; he's a RAM guy, not a CPU guy. :p:

Quote:

---

Originally Posted by EnJoY

Dude, just look at the results. Single threaded performance is barely better than that of K8. The only reason performance gains of k10 over K8 are even in the double digits are due to having double the cores and more cache.

---

Give it time man, trust me. I still remember the day Tom's Hardware tested the A64 based on singel channel 754 pin platform and believed it was no match for P4.

But we all know what happend a few months later do we now?

I hope you get what im trying to say. AMD just doesn't go about its business like Intel does. Of course Intel is currently the leader when comes to CPU performance no doubt about that. But, you lack the understanding how AMD offers its products in the retail market. Give it time , that's all im saying. Wait for the 45nm revisions and you will see how well it compares to c2d. *(Not counting the nahalem) , now thats a different ball game which is way too early even for me to speculate.

Quote:

---

Originally Posted by JumpingJack

I disagree, looking over mAJORD data, pretty much in agreement with the reviews that did similar, we are looking at 4 to 18% IPC improvement over K8 -- This is healthy, the K10 design rev from that point of view is successful.

If you compared Conroe (Core 2) over Yonah (Core) you get the same kind of 4 to 20% ish type of improvements depending on the app... any design that achieves ~ 10-15% (on average) improvement IPC wise is a good. What made Conroe shine so much was not the improvement over the prior tech along it's same lineage, but the improvement over the segment product which was along a different lineage... then again, it wasn't too difficult to improve over Prescott.

---

QFT.

And for the records. I'm sure all you guys agree A64 was a success, wasn't it?

Well, I've recolected benches results from two reviews (hardware.fr web, in this case). First, A64 launch (sep-2003, comparing FX-51 vs XP 3200+, both at 2.2ghz); second, Phenom launch (nov-2007, comparing Phenom 9600 X2, two cores vs. A64 X2 4400+, both at 2.3ghz):

http://img209.imageshack.us/img209/35/hardwarefrcc3.png

Well, Phenom perfomance gain is just fine. K10 arch is OK, IMHO.

What's been the problem? two word: clock frequency.

AMD has failed with mhz, not arch per se (on the contrary, to me arch is fine and elegant), but clock frequency, for whatever reason, some L3 bug, or MB bug, or whatever. It limits right now to launch higher freqs. But it seems they will solve it soon (with B3). And additionally, it seems that Phenom will gain even some more % perf with it :)


BTW! great work mAJORD! :)

[QUOTE=PetNorth;2598504]QFT.

And for the records. I'm sure all you guys agree A64 was a success, wasn't it?

Well, Phenom perfomance gain is just fine. K10 arch is OK, IMHO.

What's been the problem? two word: clock frequency.

AMD has failed with mhz, not arch per se (on the contrary, to me arch is fine and elegant), but clock frequency, for whatever reason, some L3 bug, or MB bug, or whatever. It limits right now to launch higher freqs. But it seems they will solve it soon (with B3). And additionally, it seems that Phenom will gain even some more % perf with it :)


Now thats really the crust of the matter now aint it? Everybody here goes by more is better and instant gratification theory and that just aint true in life in general

Quote:

---

Originally Posted by PetNorth

Well, Phenom perfomance gain is just fine. K10 arch is OK, IMHO.

What's been the problem? two word: clock frequency.

AMD has failed with mhz, not arch per se (on the contrary, to me arch is fine and elegant), but clock frequency, for whatever reason, some L3 bug, or MB bug, or whatever.

---

Architecture is definitely a problem, seeing as how as the E6600 and Q6600, for the same hardware.fr benchmarks have nearly the same performance advantage (even when clockspeed is accounted for) over Phenom as the Phenom has over the Athlon X2. So, AMD needs to clock at least 10-15% higher than Intel to be competitive but given the current power consumption, doesn't have the thermal headroom to do so even if the core can scale that high.

With proper cooling I don't see the issue if thats the case:clap: As I am sure on Phase you can crack that whip and make that Biatch sing:hrhr:

Quote:

---

Originally Posted by PetNorth

QFT.

And for the records. I'm sure all you guys agree A64 was a success, wasn't it?

Well, I've recolected benches results from two reviews (hardware.fr web, in this case). First, A64 launch (sep-2003, comparing FX-51 vs XP 3200+, both at 2.2ghz); second, Phenom launch (nov-2007, comparing Phenom 9600 X2, two cores vs. A64 X2 4400+, both at 2.3ghz):

http://img209.imageshack.us/img209/35/hardwarefrcc3.png

Well, Phenom perfomance gain is just fine. K10 arch is OK, IMHO.

What's been the problem? two word: clock frequency.

AMD has failed with mhz, not arch per se (on the contrary, to me arch is fine and elegant), but clock frequency, for whatever reason, some L3 bug, or MB bug, or whatever. It limits right now to launch higher freqs. But it seems they will solve it soon (with B3). And additionally, it seems that Phenom will gain even some more % perf with it :)


BTW! great work mAJORD! :)

---

Thanks, and great work yourself with that table. highlights the point's made.

It's important to note also that infact the gains from K8-K10 on average are probably MORE than K7-K8 if you look purely at CORE enhancments., esp those resulting in some huge (and much needed to compete with core) SSE performance boosts.

Te large gains in gaming + Archiving in the K7-K8 comparison are largly due to the move to IMC + hypertransport. Much as that was the focus of the A64 arch, native quad and enhancing SSE performance were the biggest priorities this round.



More benches to come today. Been running Wprime 1.55, X264 encoding bench, Pov-ray, Systemtool

Quote:

---

Originally Posted by accord99

Architecture is definitely a problem, seeing as how as the E6600 and Q6600, for the same hardware.fr benchmarks have nearly the same performance advantage (even when clockspeed is accounted for) over Phenom as the Phenom has over the Athlon X2. So, AMD needs to clock at least 10-15% higher than Intel to be competitive but given the current power consumption, doesn't have the thermal headroom to do so even if the core can scale that high.

---

Architecture is not the problem. K10 is quite a bit more efficient than K8. In fact, it's equally as Core is than NetBurst. AMD needs to scale clock speeds the way they planned to. In my opinion, they should not have released Phenom X4 until they had ironed out production issues and errata.

Quote:

---

Originally Posted by tux08902

Architecture is not the problem. K10 is quite a bit more efficient than K8. In fact, it's equally as Core is than NetBurst. AMD needs to scale clock speeds the way they planned to. In my opinion, they should not have released Phenom X4 until they had ironed out production issues and errata.

---

I think Accord99's point was that the architectural revision did not yield a per clock advantage over the Core microarchitecture.

nice results! thanks for posting all thoe results :toast:

Now I think:
The problem ain't that K10's gains IPC-gains over K8 are bad(the sole problem compared to K8 is the clockspeed) or that it isn't a good cpu-arch. However Intel's perfect execution and a good arch on a lower process node *is* a serious problem.
thank you for the results so far!

Increasing default HTT may have helped also, i see no reason why we could not have had 333 default HTT over 200...many are trying to test this as reports are saying it speeds up L3 thru put, the only issue is HTT over HT3 is very limited at this time.

Quote:

---

Originally Posted by Tony

Increasing default HTT may have helped also, i see no reason why we could not have had 333 default HTT over 200...many are trying to test this as reports are saying it speeds up L3 thru put, the only issue is HTT over HT3 is very limited at this time.

---

Good point!

But even if we would want to set HTT to 333 using HT2.0 we can't do it through BIOS on AM2+ boards. I've tried on GA-MA790FX and everytime I set 1GHz link or belove = no POST :shrug:

on my weird Gigabyte mobo OC with K8 - 400MHz HT, with Phenom only 230MHz ... thats strange

Not strange if you know whats going on...working with motherboard vendors now to add NB multi adjustment to bios, this may allow upwards 230HTT.

Quote:

---

Originally Posted by Tony

Not strange if you know whats going on...working with motherboard vendors now to add NB multi adjustment to bios, this may allow upwards 230HTT.

---

It doesn't with me. I've had HT and NB to 1x and 5x or even to 10x and 11x and the results are the same. Not 230HT, that's very low, but 266-268HT.

Really great job! Congrats!

Can I ask for some sciencemark and f@h benchies? :)

mAJORD, great work, thanks for all the hard work and patience with the trolls

wanted to ask a big request, can you disable 2 of the cores and run the same tests to see how it truly compares to dual-cores, or even single-cores..., and if you really want to try something, try disabling different combinations of cores to see if there are cores that work better with others. More important would be at least disabling 2 cores. If you do so, you'll have my gratitude, as well as gratitude from many here. :D

Also agree this is great work mAJORD....not every day someone takes hardware and objectively runs so much stuff on two setups :yepp:

And the differing settings don't bother me so much...the 1-2% isn't a big deal (though exact settings isn't a bad thing :)).

But I'd also agree that forcing the K10 into 2-core mode would be a very good thing to do....and it will open up the options for various benchmarks. 3D06 CPU score, WinRAR MC, Cinebench MC, etc, will all be fair game :)

The easiest way, IMO...is to go into start -> run -> msconfig (hit enter) -> BOOT.INI tab -> Advanced Options -> check /NUMPROC= -> 2 -> close and restart :)

Allows any program to open any number of processes it wants and still limits them to 2 cores being used. Also doesn't allow GPU drivers to go on the other cores either...something that may effect 3D06.

Can't wait for more tests, esp video encoding and SSE stuff :up:

I'm watching this thread with great interest... looking at quad core upgrade... not sure to dump E6850 & X38 mobo... :D

The thing that will probably move me toward Penryn is not the clock for clock speed difference and rather the 1k overclocking advantage it will likely have on air. (3.0GHz vs 4.0Ghz) I won't even get into more elaborate cooling.

Still, I might build another rig with a 9600 BE anyway.

Hey Guys, Sorry I haven't updated this. some IRL stuff has come up , including having to fix my car!

I've got results for more stuff just need to put them up

SmartAss, hopefully having a little LAN here so should be able to get a bunch of games on it off a friend. So that'll be nice.

will phenom x2's oc well do you think?

if so, that should start to displace the k8 x2's conclusively i wooda thought.

tanxs for the results.

I did some tests for you guys simulating Phenom X2 2200MHz with DDR-800 4-4-4:


http://img410.imageshack.us/img410/6...ueenxo3.th.png
CPU Queen 4395 vs 4262 of X2 4200+ 2.2GHz

http://img410.imageshack.us/img410/1...orxxtl8.th.png
CPU PhotoWorxx 26375 vs 17902 of X2 4200+ 2.2GHz

http://img410.imageshack.us/img410/2...zlibgf7.th.png
CPU ZLib 29487KB/s vs 26238KB/s of X2 4200+ 2.2GHz

http://img410.imageshack.us/img410/3...uliars6.th.png
FPU Julia 4660 vs 3742 of X2 4200+ 2.2GHz

http://img135.imageshack.us/img135/2...ndelea5.th.png
FPU Mandel 2308 vs 2114 of X2 4200+ 2.2GHz

http://img135.imageshack.us/img135/3...uliasf7.th.png
FPU SinJulia 5671 vs 5872 of X2 4200+ 2.2GHz

http://img530.imageshack.us/img530/3...erpigl4.th.jpg
Cinebench 2003 x64 + SuperPi

oh, vs the built-in results for s939 x2 4200 with ddr400

julia tests single precision floating point (32 bit), mandel tests double precision floating point (64 bit), sinjulia tests extended precision floating point (80 bit) afaik... k10 loses to s939 in extended precision (exponential and trigonometric functions according to my everest help file)? :shrug:
i bet extended precision trigonometric and exponential functions don't get used very often :p

nb speed 1.8 ghz?

Quote:

---

Originally Posted by hollo

oh, vs the built-in results for s939 x2 4200 with ddr400

julia tests single precision floating point (32 bit), mandel tests double precision floating point (64 bit), sinjulia tests extended precision floating point (80 bit) afaik... k10 loses to s939 in extended precision (exponential and trigonometric functions according to my everest help file)? :shrug:
i bet extended precision trigonometric and exponential functions don't get used very often :p

nb speed 1.8 ghz?

---

Yes, NB 1.8GHz is default for Phenom 9500!

Slower SinJulia results might be caused by x64 OS.... By default you can't use 80bit FP, only 64/128bit... I will re-run this test under Win XP MCE to confirm that... :)

With these new boards that can do over 1000HTT, would we see any performance increase on an K8 by using say 2000HTT instead of the stock 1000HTT? Can someone do a quick test? Thanks. Great thread btw.

Quote:

---

Originally Posted by awdrifter

With these new boards that can do over 1000HTT, would we see any performance increase on an K8 by using say 2000HTT instead of the stock 1000HTT? Can someone do a quick test? Thanks. Great thread btw.

---

My DFI 570 ultra did 1600 HTT easy, prime stable. I did run 8*400 for a while but did not find any preformace leaps over my 6400+´s stock 16*200.

Hey Guys I've updated the first Post with new benchies

Lightman: Thanks for your Additions, do you want me to paste that into the main thread under your name with a link to post?

Thanks mAJORD :).You have done an amazing job :).

edit:i saw you 've seen Lightman's post about GB "bios good news" :)

For everyone else,Lightman managed to break 240HTT limit for his Phenom and GB board.Post here

Thanks for the sicencemark results!

Found it strange to see that the results points toward equivalent performance in everything, except blas performance. Strange behaviour, any clues?

Can I ask for an F@H bench please? Thanks a lot! :D

And really good work, that seems to be the best job until now on such a subject. Do you plan to redo everything with B3 if it starts to show strong differences in performance from B2?

Quote:

---

Originally Posted by BadNizze

My DFI 570 ultra did 1600 HTT easy, prime stable. I did run 8*400 for a while but did not find any preformace leaps over my 6400+´s stock 16*200.

---

Cool, thanks for the test.

Very nice tests !!

To bad that Phenom's clockspeeds cant get that high like the 3200 of the K8. Hopefully AMD can fix dit and push phenom above 3200 within 120 watt TDP.

Quote:

---

Originally Posted by mAJORD

Hey Guys I've updated the first Post with new benchies

---

Thanks!

I believe there is 2 small mistakes. For Systool, the results are invert:
In fact from your screenshoots:
Phenom: 237sec
Athlon: 263sec
So K10 + 10.9% is still valid


But the other about x264 Benchmark:
I don't see time on your screen, but fps seems to shows that time you give after is good.

Phenom:
PASS1:89.00
PASS2:17.58

[106.58]

Athlon:
PASS1:75.13
PASS2:17.97

[93.1]

So it should be K10 -14.4% overall (in this test K10 underperform K8)

Quote:

---

Originally Posted by mAJORD

x264 Benchmark

Phenom:
PASS1:89.00
PASS2:17.58

[106.58]

Athlon:
PASS1:75.13
PASS2:17.97

[93.1]


K10 +14.4% overall


Systool
Phenom: 263sec
Athlon: 237sec

K10 + 10.9%

---

@nemrod

How can K10 underperform K8 when it has a lot more fps in x264 test?

Interesting, great work, thanks. :toast:

I did a X2/C2D/X4 comparison at similar MHz (2640MHz) but I was too short of time to replicate much and the Phenom was giving me NB problems to continue on. Plus I couldnlt get 2k6 to run on the C2D build no matter what.

I might post my results in my K9A2 thread later, but I don't see the point since you have already done most of it.

Problem is, if you try and get a high NB fre. like X2 has you start seeing many similarities in perf.

Quote:

---

Originally Posted by informal

@nemrod

How can K10 underperform K8 when it has a lot more fps in x264 test?

---

I was thinking it was time for compression. For fps, I was looking at the second pass. But just an error
:shrug:

Anyway, this benchmark seems dependent of the number of core:
http://www.techarp.com/x264_Benchmar...ts_amd_big.gif