Sign in with
Sign up | Sign in
Your question

Q6600 system vs AMD Phenom system - Page 3

Last response: in CPUs
Share
July 1, 2008 8:22:43 PM

I think this with I/O is a bit to advanced for some of the participants here.
a b à CPUs
July 1, 2008 8:28:34 PM

man guys, kassler is prob a 45 year old that lives in his grandmothers basement. just ignore him and he will go away
a b à CPUs
a b V Motherboard
July 1, 2008 8:31:33 PM

kassler said:
I think this with I/O is a bit to advanced for some of the participants here.

No, I think you don't understand the idea that something other than the CPU could be bottlenecking the system.
Related resources
July 1, 2008 8:39:36 PM

OK so I am to blame for this since I asserted that the Phenom is slower than the Q6600. I got this idea from viewing Tom's charts. Other than that in my house we have 3 desktops and use AMD Athlons and 64 X2 - because they have always been very fast CPUs and quite cheap.

Now as a matter of interest, at work we are switching out of the AMD platform and into Intel, and so are other big places. We are using the new (?) Xeon quads and they beat the 4-way and 8-way Opterons. I am not 100% certain on memory / cache / coherence mechanisms, since I was not able to run a test like for like (ie 8-way AMD vs 8-way Intel) and am quite suspicious of the 4-core Intels, do they suffer cache contention perhaps? However on per-CPU performace the 2.33 GHz Intel Xeon quad beats the 2.66 GHz Opteron by about 40%. You may ask, "what kind of test was it and was it trying to encode 16 GB of RAM" ? No, it was hashing a 64KB string repeatedly as well as populating a 64K string, as well as performing a varietry of assembly code found in C++ applications. We also run another test harness which , outside of raw processing also accesses MB of RAM and again the Opterons were left behind. Obviously the Inbtel platform is also available in 2.66 GHz and 3.0 GHz stock clock speeds, which means there is no AMD equivalent since the quickest AMD (the 2.66 GHz) is slower than the 2.33 GHz Intel. Which is a reversal of roles because in the past it used to be that AMDs were faster than Intels per clock cycle.

July 1, 2008 8:47:41 PM

akistzortzis said:
Now as a matter of interest, at work we are switching out of the AMD platform and into Intel, and so are other big places. We are using the new (?) Xeon quads and they beat the 4-way and 8-way Opterons. I am not 100% certain on memory / cache / coherence mechanisms, since I was not able to run a test like for like (ie 8-way AMD vs 8-way Intel) and am quite suspicious of the 4-core Intels, do they suffer cache contention perhaps? However on per-CPU performace the 2.33 GHz Intel Xeon quad beats the 2.66 GHz Opteron by about 40%. You may ask, "what kind of test was it and was it trying to encode 16 GB of RAM" ? No, it was hashing a 64KB string repeatedly as well as populating a 64K string, as well as performing a varietry of assembly code found in C++ applications. We also run another test harness which , outside of raw processing also accesses MB of RAM and again the Opterons were left behind. Obviously the Inbtel platform is also available in 2.66 GHz and 3.0 GHz stock clock speeds, which means there is no AMD equivalent since the quickest AMD (the 2.66 GHz) is slower than the 2.33 GHz Intel. Which is a reversal of roles because in the past it used to be that AMDs were faster than Intels per clock cycle.


9 of 10 companies wants performance using databases when it comes to server computers. there arent that many companies that need computation performance in their servers. But if you run tests that L2 cache is enough for then intel is very fast
July 1, 2008 9:03:40 PM

You use I/O to send millions of polygons to the card? Do you not use its main RAM? Anyway, in which games/applications (not synthetic tests) are the Phenoms faster than the Q6600 ? And what is the % of those games/applications?

For example there are some CPU charts at Tom's where the dual cores beat the quad cores, but that means the application tested did not make good use of the 4 cores (both Intel and AMD) - at work our applications are multithreaded and are written to scale across all cores/CPUs. We also care for latency which means we also care of the absolute speed of each core because there are some things that cannot be split across multiple cores. That is for example why I was reluctant to replace my Athlon 64 X2 Dual Core 6000+ with the Q6600, because some applications may rely on a single thread for doing the work and there the 3.0GHz Athlon may beat the 2.4 GHz Q6600. However seeing how easy it is to overclock the Intel to 3.0 GHz and beyond, and having abysmal luck trying to overclock the Ahtlon, I am not worried anymore. The only small hurdle is to justify the expense of 1000 pounds for a new PC when the "old" one is actually not bad at all :) 
July 1, 2008 9:03:50 PM

kassler said:
check the game tests


Oh you mean Crysis?

oops, no. Phenom got whipped right there.

Oh I know, you must mean Knights of the sea?

oops, no. Phenom again, got pwned right there.

Same with WIC. Same with CoD4.

The only game Phenom shined was Company of Heros.

Still no proof on your BS that "I/O performance matters".
July 1, 2008 9:10:00 PM

It is astonishing that there are so many that don’t understand the test that I linked to. I am starting to understand why it is problematic to sell something that you need to be more skilled to understand.
July 1, 2008 9:12:30 PM

I am also starting to understand why THG's quality is sinking every second...
July 1, 2008 9:15:41 PM

yomamafor1 said:
Oh you mean Crysis?

oops, no. Phenom got whipped right there.

Oh I know, you must mean Knights of the sea?

oops, no. Phenom again, got pwned right there.

Same with WIC. Same with CoD4.

The only game Phenom shined was Company of Heros.

Still no proof on your BS that "I/O performance matters".


You are mixing up I/O with processor performance. Intel is a faster processor, amd has better I/O. I/O is not that important when the resolution is low. When resolution is low hardware isn't that important. Your monitor will not show over 60 fps so it doesnt matter that the computer can show 600 fps ;) 
July 1, 2008 9:22:36 PM

Buggering up yet another thread, Assler?

You still don't get it and you probably never will. I/O performance just doesn't make a huge difference in desktop apps the way it does in server apps. When you come into a thread and make unsubstantiated claims, you are expected to post some sort of evidence to back up those claims. The evidence you post should also be relevant to those claims. Basically, you are convicting a man of murder because his brother living 100 miles away yelled at someone. It's not at all relevant to the case, but you're trying to present it as evidence anyway.

Please post RELEVANT information. Yes, AMD has the edge in I/O performance, but that's quite obviously not enough to be the top performer on the DESKTOP. You can go on and on and on (and you obviously have) about Intel vs. AMD scalability until you're blue in the face... but if it's not relevant, then who the hell cares? When it DOES make a difference, that's when the rest of us will care. You have yet to show us any compelling evidence to make us switch from Intel to AMD right now.
July 1, 2008 9:41:59 PM

zenmaster said:
Depends on the Application to see the amount of scaling.
Some apps scale nearly 100%.
Others do not.

There are MANY reviews that show that increasing FSB and other Memory settings that increase bandwidth have little to no effect on performance.

Again, Please Link Reviews from a Well Trusted Website that shows the Intel Q6600 performing poorly on multi-threaded apps compared to the Phenom.

There are already links in this thread that show that the Intel does not have any issues.

There are also many OCing tests that show the Q6600 doing increasingly well as it overclocks in mult-threaded apps. If there was a memory wall that would not be the case.

The Q6600 will not hit a memory wall.
The Phenom can't compete, unless you can't the fact I've seen the Phenom 9600 for under $100 as part of some combos.



This is engineering simulation.

At such low thread numbers scaling is virtually linear.

I'm not sure the data I have here is meant to be public domain, and it is not from a website.


Trust me, the kentsfield and yorkfield processors suffer from severe cache contention and bandwidth contention for CFD - for the majority of desktop applications the cache size is sufficient to keep the compute cores busy, for CFD it falls far short. Its a feature of the architecture that will be fixed when Nehalem arrives.



If your dying for an indication from a 'reputable' website:

http://www.intel.com/performance/server/xeon/hpcapp2.ht...

speed up from dual core, 3.0 GHz, 4MB L2 cache FSB1333 to quad core 3.0 GHz, 2x4MB L2 cache FSB1333 = 1.5, not the 2 you would expect - simply due to contention across the FSB (never mind within the cache itself).

If Intel had run that on CPU 0/1 and CPU 2/3 of the quad core, the difference from 2 threads to 4 would have been even less as each thread would have full access to the 4MB L2 cache on its individual die.



Phenom does not have that problem - sure, its L3 is slow, but main memory access is quick enough to keep the procs running flat out.
a b à CPUs
a b V Motherboard
July 1, 2008 9:46:28 PM

Very true for CFD work (though not quite as bad with 1600FSB), but that is atypical of most desktop applications. The argument with Kassler is not that there is no application that is limited, but that most common ones are not, including games (even at high resolution). I'm waiting until Nehalem until I build a desktop anyways, but the main reason is not the new memory architecture.
July 1, 2008 9:46:35 PM

kassler said:
You are mixing up I/O with processor performance. Intel is a faster processor, amd has better I/O. I/O is not that important when the resolution is low. When resolution is low hardware isn't that important. Your monitor will not show over 60 fps so it doesnt matter that the computer can show 600 fps ;) 


Again, you have 0 clue what you're talking about.

When resolution is low, the burden is on the CPU. That's why almost every review sites use this method to test in game performance. The "I/O" is also the greatest at low resolutions.

However, as resolution goes up, the bottleneck slowly shifts towards GPU, and GPU performance becomes more of a deciding factor. So that's why you see as resolution goes up, the difference between test systems decreases.

It also doesn't matter if your monitor can display 60 FPS or 600FPS. Almost no one plays at 800x600. Therefore that resolution is more practical to use as a benchmark resolution than real world gaming resolution.
July 1, 2008 9:52:35 PM

kassler said:
It is two different processors built for different tasks. If you are going to compare the phenom with a intel then a better comparison would be the next Nehalem, those are very similar processors.
AMD is built to scale, intels (core) is built to run one single thread as fast as possible. Quad’s are just two duals glued together.
It is much more complicated to build a processor for scaling, it isn’t possible to create a processor that can work faster running single threaded applications faster than one processor built for that purpose.
Some multithreaded applications is fast on intel, if each thread works as one single thread (don’t talk to other threads and is big) in the multithreaded application.

You don't need to overclock one AMD, you need to overclock Intel.

Intel has a slow FSB that handles all memory traffic and all I/O traffic.
Phenom has Hypertransport 3.0 for I/O (~20 GB/s and it doesn't need to handle memory)
The FSB on intel is about ~10 GB/s, and it isn't that effective
I just got back and this thread has wandered around a little. I just thought I would re-post the initial post from kassler that pissed me off in the first place. You won't be asking to compare the Phenom to the Nehalem with the IMC when it comes out. Your arguments about it comparing to the trusty double cheeseburger Q6600 @ 3G in single and multithreaded desktop apps are BS.

We all just have to hope that AMD doesn't piss off Intel and cause them to bring their full weight to bear.
July 1, 2008 9:53:49 PM

cjl said:
Very true for CFD work (though not quite as bad with 1600FSB), but that is atypical of most desktop applications.


No and yes.

The FSB1600 is quite bad too - from 4 procs ideally used (CPU0, 2, 4, 6) you can see a speedup of 3x (compared to 1 cpu), however increasing to 8 procs your speedup only goes to 3.2x. :( 

Unfortunate for us at the moment, but hey...


But yes, CFD definitely is not a typical desktop application.


Quote:

The argument with Kassler is not that there is no application that is limited, but that most common ones are not, including games (even at high resolution). I'm waiting until Nehalem until I build a desktop anyways, but the main reason is not the new memory architecture.


If the L2 cache of a Conroe is large enough for immediate CPU data requirements, then a Conroe will beat a similarly clocked Phenom quite easily.

I don't think anyone can dispute that really.

July 1, 2008 10:12:56 PM

kassler said:
You are mixing up I/O with processor performance.


No, he's not. He's asserting that I/O does not matter(because neither AMD or Intel is saturated at the desktop level), but PROCESSOR PERFORMANCE DOES.


REAL RESULTS vs. artificial memory benchmarks


What do you want? I'll take real results.
July 1, 2008 10:30:56 PM

Keep arguing with him. . .

all the top minds of your forum will leave and form their own forum site.






a b à CPUs
a b V Motherboard
July 1, 2008 10:34:22 PM

Amiga500 said:
No and yes.

The FSB1600 is quite bad too - from 4 procs ideally used (CPU0, 2, 4, 6) you can see a speedup of 3x (compared to 1 cpu), however increasing to 8 procs your speedup only goes to 3.2x. :( 


Yes, but it isn't too bad for a single socket system, even on CFD (3x with 4 cores isn't half bad). Anything multi socket with those kind of apps, and you want Opterons, no question.
July 1, 2008 10:46:27 PM

jkflipflop98 said:
Keep arguing with him. . .

all the top minds of your forum will leave and form their own forum site.


That would never happen.

:heink:  :sweat: 

:lol: 

sssssh.... :non: 
July 1, 2008 11:42:52 PM

jkflipflop98 said:

all the top minds of your forum will leave and form their own forum site.


Yes... that could happen...

but then the Intel fanboys would just find us and it would become the same as this site.
July 2, 2008 12:05:34 AM

Maybe it's not the Intel fanboys. Did you ever consider that? :lol: 
July 2, 2008 12:08:09 AM

jkflipflop98 said:
Keep arguing with him. . .

all the top minds of your forum will leave and form their own forum site.



Sorry, that already happened.

(although I know it was tongue in cheek)
July 2, 2008 12:15:00 AM

Actually, it's people like Baron Matrix, Assler and yourself Keith that caused a great split not so long ago.
July 2, 2008 12:24:16 AM

Zoron said:
Actually, it's people like Baron Matrix, Assler and yourself Keith that caused a great split not so long ago.


Nah, it was mostly BM, not kassler or keithlm, although they are trying imitate BM as much as possible. But BM was in a reality all his own.
And don't worry, all those who left, are doing fine and dandy.
Except BM, cause I haven't heard from him since the folks at AMDZone suposedly ripped him a new one. LOL.

:lol: 
July 2, 2008 1:00:18 AM

kassler said:
It is astonishing that there are so many that don’t understand the test that I linked to.
A vast majority of people seem to be disagreeing with you. You say there are many who don't understand, but with the tide of opinion and benchmark data being in such overwhelming disagreement, have you considered that, just possibly, there's only one who doesn't understand?

There is no greater barrier to truth than the presumption that you already posess it.
July 2, 2008 1:02:17 AM

Zoron said:
Actually, it's people like Baron Matrix, Assler and yourself Keith that caused a great split not so long ago.


Does it make you feel superior to call people by childish nicknames? Do you know what this says about you psychologically?

Do you understand that personal attacks are generally considered "poor" etiquette in forums? You can not further your opinion by calling people names.

July 2, 2008 1:50:25 AM

keithlm said:
... personal attacks are generally considered "poor" etiquette in forums? You can not further your opinion by calling people names.
+1
July 2, 2008 3:17:34 AM

keithlm said:
Does it make you feel superior to call people by childish nicknames? Do you know what this says about you psychologically?

Do you understand that personal attacks are generally considered "poor" etiquette in forums? You can not further your opinion by calling people names.



Wah.

Seriously, if you're going to post in a public forum, you have to have a bit of a thick skin. I simply call things the way I see them... political correctness is utter BS to me. If he's going to act like an @$$ and not listen to opposing viewpoints and keep shoving the same old "evidence" down our throats... then I'm going to continue to call him "Assler". It's an attention-getting device... since he seems to ignore everything anyone else says. Personally, I'm sick of hearing the same argument over and over... repeating it won't cause it to magically gain credibility. I acknowledge AMD's edge in server and memory I/O intensive applications... however, he still completely ignores every fact that everyone has presented showing him that Intel still holds sway in normal desktop applications. Until he can admit that Intel does in fact beat AMD in most cases on the desktop, he will continue to be the @$$ I perceive him to be...
July 2, 2008 4:12:29 AM

Right now, until Deneb arrives, the only worthwhile Phenom is an 8750 on a 780G board for a budget system or HTPC. If you're building now, I'd recommend Intel. Too bad there are no longer any AMD/ATI chipsets for newer Intel CPU's. Still have a P4630 on an ATI X200 ASUS board for the legacy 4th PC.

Afer reading the reviews of the new Phenom 9950 @ 2.6 with thermals of 140, I'm even more convinced that AMD fans need to wait and those on the fence won't be hurt by going Intel.

Just make sure to get a good AMD GPU. That is one area where they shine in this generation.
July 2, 2008 4:16:26 AM

More to the point of the topic: there is practically nothing you could do which will make a Phenom beat a C2Q in the applications where a C2Q is strong.

AMD made a mistake, in as much that they designed a quad core processor that is great for server apps but only just fair for desktop apps - but they meant to make it for desktops!

There is nothing inherently all that bad about Phenoms, especially the 9850BE - thats actually a pretty solid processor for 235$ - but I guess the real problem is that it still gets outperformed by its Intel counterparts and the Phenoms are still pretty expensive to produce in comparison which means a poor margin for AMD.

Lets hope AMD's next architecture is better.
July 2, 2008 6:40:25 AM

Are we really talking Hypertransport vs FSB ?
a b à CPUs
a b V Motherboard
July 2, 2008 7:04:37 AM

Well, that's part of it. More whether that matters more or whether the core matters more.

Basically, whether mediocre interconnect + amazing core and cache (Core 2) is better than amazing interconnect + mediocre core (phenom)
July 2, 2008 9:23:27 AM

ovaltineplease said:
More to the point of the topic: there is practically nothing you could do which will make a Phenom beat a C2Q in the applications where a C2Q is strong.

AMD made a mistake, in as much that they designed a quad core processor that is great for server apps but only just fair for desktop apps - but they meant to make it for desktops!

There is nothing inherently all that bad about Phenoms, especially the 9850BE - thats actually a pretty solid processor for 235$ - but I guess the real problem is that it still gets outperformed by its Intel counterparts and the Phenoms are still pretty expensive to produce in comparison which means a poor margin for AMD.

Lets hope AMD's next architecture is better.


I'm not sure you got that right. The Phenom is based on the Barcelona. AMD develops server processors. Everything else they sell or produce in the CPU market is derived from their server branch. The Phenoms are probably mostly processors that didn't cut it in the server department because of broken HT links or too high power requirements. Just like the Toliman - you can't sell a quad with a broken core to a server farm.


edit: I've to mention that AMD needs to lower the prices. In Europe they are selling a Q6600 for 134€ (211$) while a 9750 costs 144€ (221$). In some countries AMD might still be competitive, but in most european countries they are not.
July 2, 2008 1:23:44 PM

"AMD develops server processors". "AMD beats Intel in the server world".

Is that right?

I have tested the following server systems at work.
ProLiant DL585 G2 (Opteron 8218 2.6 GHz, 8 cores)
ProLiant DL585 G1 (Opteron 875 2.2 GHz, 8 cores)
ProLiant DL385 G1 (Opteron 275 2.2 GHz, 4 cores)
ProLiant DL740 G1 (Xeon 3.0Ghz, 8 cores/HT)
ProLiant DL380G5 (Xeon Quad Core 2.33GHz/3.0GHz, 8 cores)
Some examples of results.
The first test is a simple CPU test which uses just 64 K data. The latest Intels are 90% faster than the AMDs. As a matter of fact the 2.33 GHz Xeon Quad beats the 2.66 GHz AMD Opteron by 25%.
I then ran throughput tests (involving lots and lots of memory) but could not get my hands on the DL585G2, only the older G1. Still the Intel Xeon Quad produced 65% more throughput. The Opteron 8218 system will be quicker than the Opteron 875 but I do not expect miracles. As it currently stands I cannot see where AMD could potentially beat Intel, desktops or servers. Its flagship 8218 is 2.6 GHz and gets beaten by the 2.33 GHz Xeon Quad, and in terms of memory, bandwidths and all that, the Intel still comes easily tops. And with Intel you can get the 3.0GHz (or more?) Xeon Quads that go even faster.
a b à CPUs
a b V Motherboard
July 2, 2008 3:35:18 PM

There are certainly applications in which AMD does work better, especially in quad socket systems.

Here is a good example: http://it.anandtech.com/IT/showdoc.aspx?i=3335

Yes, the Intels are still best for many things, but there are a couple where AMD performs quite nicely in the 4 socket systems.
a c 127 à CPUs
a b V Motherboard
a b À AMD
July 2, 2008 3:40:11 PM

AkisTzortz is

In single CPU and sometimes in 2P servers, Intel does quite well. But in 4P and higher servers, AMD tends to do better since the IMC and HTT links scale better than the FSB and the FSB does tend to get saturated with more CPUs.

Of course that will change with Nehalem but when it comes to the desktop market the FSB does not get saturated and Intel performs the best. But others will try to change your mind because they only see loyalty. I myself am only loyal to one thing really. I try to buy American products, especially cars.

But in the end its all about what the AMD fanatics think is right. God I hate when they try to push crap on ya.
July 2, 2008 4:03:30 PM

jimmysmitty said:
In single CPU and sometimes in 2P servers, Intel does quite well. But in 4P and higher servers, AMD tends to do better since the IMC and HTT links scale better than the FSB and the FSB does tend to get saturated with more CPUs.


I agree with that statement.
July 2, 2008 5:45:54 PM

akistzortzis said:
"AMD develops server processors".

Is that right?


Yes.
July 2, 2008 7:00:40 PM

kassler said:
I think this with I/O is a bit to advanced for some of the participants here.



I/O = Input/Output

Input = mouse, keyboard, lightpen, scanner

output= monitor, printer, speakers...

if what the amd fanboy is saying is correct, we'll be using 6 keyboards, 4 mice, and prolly 10 monitor on each desktop in the future, hurray!!!

July 2, 2008 8:45:04 PM

wh3resmycar said:
I/O = Input/Output

Input = mouse, keyboard, lightpen, scanner

output= monitor, printer, speakers...

if what the amd fanboy is saying is correct, we'll be using 6 keyboards, 4 mice, and prolly 10 monitor on each desktop in the future, hurray!!!
No, he's talking about traffic on the bus.
July 2, 2008 9:14:10 PM

spongebob said:
No, he's talking about traffic on the bus.

The short, yellow one, to be precise.
July 3, 2008 4:37:03 AM

Slobogob said:
The short, yellow one, to be precise.

:lol: 
July 3, 2008 7:00:15 AM

Out of interest, are we saying that the Intel Quads, XEons and all, are simply two Duals sandwhiched together, and for all intents and purposes they may as well have been two sepatate chips?

To explain, are we saying that when it comes to cache coherence and memory accesses there is absolutely nothing to gain by the Intel Quads and that CORE#3 will need to go through the FSB when it needs to get some data which lives in the L2 cache of CORE#0 even though they are both on the same chip?
July 3, 2008 7:13:15 AM

That's correct. Therefore under memory intensive applications, or applications where core-to-core interconnect is crucial, you'll see Xeon losing even to the K8s.

This is also why Intel is moving on to monolithic multi-cores, with Nehalems. They didn't bother doing it due to immature process nodes, and horrendes yields, which are things AMD is currently suffering from. As 45nm process matures, it will significantly increase native quad core's yield rate.
July 3, 2008 8:20:17 AM

My test results between an 8 core AMD 875, AMD 8218 and Intel Xeon Quad (8 cores):

Conclusion:

I have observed the AMD is faster in shifting data around between cores RELATIVE TO ITS COMPUTATIONAL SPEED, but real life is not just shifting data, you also need to look at the data, and the AMD cannot catch up with the much faster Intel simply on that account.

Some numbers:

Intel Xeon Quad 3.0 GHz :
computational speed: 0.65ms
data transfer: 33% * 0.65ms = 0.22 ms

AMD 8218 2.6 GHz:
computational speed: 0.96ms (50% slower than the Intel)
data transfer: 25% * 0.96ms = 0.24 ms

From the above youc an see that even though the AMD needs 25% of its time to shift data compared to the 33% of the Intel, its baseline is slower, and combined with the fact that real life is not shifting data between cores without ever looking at it, makes it much slower.

Perhaps there are some genuine applications out there that would show the AMD 8218 to be quicker than the Intel. Or perhaps AMD make 3.2 GHz cores that might catch up the Intels?
July 3, 2008 9:16:26 AM

You certainly hit the nail on the head.

Move that understanding to the desktop and you put the nail in the coffin.

You weren't trolling when you started this thread were you?
July 3, 2008 9:24:58 AM

akistzortzis said:
My test results between an 8 core AMD 875, AMD 8218 and Intel Xeon Quad (8 cores):


You are comparing between 2 4P K8 cores and a 2P Xeon?


What are the tests?
July 3, 2008 10:34:53 AM

I guess he would need to use Phenoms to match the quad Xeons, but the ultimate conclusion is correct, or am I in left field?
July 3, 2008 12:28:17 PM

Zorg said:
You certainly hit the nail on the head.

Move that understanding to the desktop and you put the nail in the coffin.

You weren't trolling when you started this thread were you?


No. I was really thinking of getting a new PC. All my other PCs in the past **many** years have been AMDs. The current PCs at home are Athlon 64 X2 6000+, Athlon 64 X2 4400+, Sempron 2600+ and an Intel Core Duo on the Dell laptop. I am not biased towards Intel, I could not care less who makes it. But Tom's charts do show the Q6600 be quicker than Phenom and the AMD is not cheaper, plus the Q6600 overclosk easily. No brainer and it was flowing smoothly until Kessler cut in.
!