Phenom users 2.....

EdzBourne

Distinguished
Dec 27, 2007
39
0
18,530
0
I have posted my exprerience in using Q6600 and ask a comparison of Phenom in MULTI-TASKING SMOOTHNESS yesterday..

the link:
http://www.tomshardware.com/forum/247803-28-phenom-users

And still people are bashing AMD...I am a Fan of either, I've used both AMD & Intel before..I don't care if Phenom is 15% slower than Q6600, cause they are cheaper also...

The question is not the speed of work it does, but the SMOOTHNESS it can multi-task, because in a multi-tasking environment, there should be a benefit of having a Cache shared by all processor than separate Cache with intel...
Just like the Pentium D 8xx & 9xx having separate cache while the Athlon X2 are shared cache...

 

turpit

Splendid
Feb 12, 2006
6,376
0
25,780
0
IF YOU DONT OWN A PHENOM, OR HAVE A LINK TO A FORUM WHERE SOMEONE DOES, STAY OUT OF THIS THREAD.

IF THIS THREAD DEGRADES INTO ANOTHER INTEL VS AMD BASH BY PEOPLE WHO DONT OWN PHENOMS, ILL START BANNING
 

homerdog

Distinguished
Apr 16, 2007
1,700
0
19,780
0
Well I have a 939 3800+ X2 and multitasking is silky smooth. Seeing as how K10 isn't much different than K8 (especially with the TLB patch in place) I would speculate that Phenom is pretty smooth as well.
 

yomamafor1

Distinguished
Jun 17, 2007
2,462
0
19,790
2
You won't see any differences in smoothness on the desktop between Phenom and Q6600. Unless you have a program that require 100% utilization from all four cores simultaneously, you won't notice anything different.

If you do have a program that takes that much CPU utilization, you're looking more to the server segment.
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780
0
EdzBourne, what exactly do you mean by smoothness?

For example, no lag, no stutter when handling multitudes of windows? If that be the case, nearly all dual-core and quad-core cpus are "smooth". It's actually more of the HD that'll cause the stutters.

Let's assume that the HD isn't a problem. Logic is that a faster cpu will be "smoother" than a slower one. Therefore the Q6600 should be "smoother".

...geez turpit you're scaring me...
 
I feel that Tom's own benchmarks speak very well for the performance of the processors. "Smoothness" is a rather subjective term. People who have a bias may be very likely to come to their own conclusion when measuring something that is subjective. I've lost faith in AMD which has turned my AMD bias into an Intel one, and my own pro-Intel bias would have me favor Intel in subjective tests, that's why I rely on objective, fact driven, tests that are measured in hard numbers.

http://www.tomshardware.com/2007/11/19/the_spider_weaves_its_web/

The cold hard benchmarks speak for themselves. Intel's offerings are simply better, even their lowest Q6600 processor bests AMD's non-released (recalled) Phenom 9700 in nearly every benchmark except artificial memory bandwidth tests that many would argue have no real world application for consumer purposes.

So in terms of smoothness and responsiveness, the conclusion I would come to is that the processor that does better in nearly all benchmarks would provide better smoothness and responsiveness.

In my most humble opinion based on the facts available, of course.
 

pat

Splendid
Jan 2, 2001
6,399
0
25,810
19


I will try to keep thing simple...

A formula one car is fast and in no way smooth. It is hard on the driver and on the car mechanic.

Same thing happen with cpu. a cpu can be fast, but that doesnt mean it is smooth. It can execute the code really fast, but can be interrupted often if the internal don't have enough registry to store information, or higher latency due to some bus with less bandwidth. Problem is, at the speed where those interrupt happen, you are not likely to feel it. The netburst architecture, with its long pipeline was often interrupted in case of, branch misprediction, in exemple, but that was not something you could actually feel, because the proocessor was running fast(in GHz).

The core2 quad and the Phenom are 2 excellent processor, no matter what others might say. The phenom has strenght and weakness just like the Core2 Quad. The core2 quad has a faster execution compared to the Phenom on a clock by clock comparaison. Especially in single thread. But, when you start to run multithreaded application, the gap between both cpu is closing as soon as the load become more important. because the Phenom has more headroom to move data because of faster HT bus that don't saturate as easily. The Core 2 Quad has less headroom because of its glued core. That's why higher amount of cache is a benefit.

So, what is smoothness? while one application may take longer time to render on a Phenom because it is a bit slower, that do not prevent other application to move data freely. If you are rendering with a fastest CPU, it may be possible than the other app may slowdown because their data are limited in speed by lesser buses. It doesnt mean the CPU is not fast, it is simply that the subsytem is not fast enough for the cpu to work at full capacity. I'm not saying that the Core2 Quad is lagging... I'm just trying to explain why some processor are executing some program faster than another and sometime slower that the other. Because there is not only a cpu core in a computer..

To be put at a slower scale, imagine running 4 instance of Winrar. Each of them is taking data on one slow ATA33 HDD, processing it, and writing it on the same drive. And all the core has it own file to work with. Poor HDD... Do you think that the HDD will be able to hold on and feed the necessary data to each core?

So, to answer your question, yes, the phenom can be smoother when running multiple apps than a Core2, even if it takes more time to process the data internally.

And yes, I do own a Phenom. And yes it is smooth. And no, I did not get bugged by the TLB bug. And yes, even at 2.2 GHz, thing are hapenning faster than my X2 4800+ at 2.4 Ghz. But yes, sometime it fell slower too. Yes, I got an ECS motherboard. And no, I did not have any problem having it recognize my 4 gigs of RAM. And no, I did not reinstall Vista (same chipset upgrade). it started right on, installed new driver and keep up running strong since 7 days now.
 

Amiga500

Distinguished
Jul 3, 2007
631
0
18,980
0
IF you want to do a comparison, you need an objective way of doing it.

Does anyone have any repeatable scientific techniques for measuring smoothness? (or can anyone suggest any?)



In my opinion:

It (smoothness) is a measure of multi-tasking, so we might want to consider loading the 4 cores with 4 different programs, like say, a pdf converter/writer, a winrar encode/decode, a video encode and an audio encode.

Then measure time to initiate all 4 programs, which is a (bad) measure of responsiveness under increased loading, and measure time to complete all 4 programs.

Of course, the actual programs would have to be decided.



Another alternative would be to have a number of different jobs to run, say 7 - so as soon as one is finished you can start another - how quick the system lets you get it started would be a measure of responsiveness.




I have a Q6600 - and I have to say the difference between 2 and 3 cores in terms of responsiveness (when doing heavy multitasking) is amazing!

I also have a K8 X2 (4200 I think), and it was a vast improvement over the FX-53 I previously had in the box. But even it cannot match the Q6600 (when using only 3 cores - I can load one out 100% with a CFD program, it might as well be a tri core then). Anywayz, point is - if that jump from 2 to 3 cores is not specific to the programs or workload I have, AMD might have winner in the Toliman core.
 

yay

Distinguished
Jan 9, 2007
324
0
18,790
1
WOOT for turpit!!!

anyway, unless you can see at 11 billion FPS (humans see anywhere from 40-80) you will not see any difference.
 

quantumsheep

Distinguished
Dec 10, 2005
2,341
0
19,790
1
I can just say that the poster somewhere above who was saying that AMDs processors catch upto Intel's CPUs due to the faster HTT bus is wrong. If you've looked at any articles that examine Intel's FSB you will see that even at 1066 there is little, if any, bottleneck on the Intel CPUs.

The "performance increase" gained by having HTT was more of a marketing gimmick my AMD than any tangible performance increase. Sure FSB is getting towards the end of it's life, but it is not quite saturated yet. Especially with the new Intel CPUs cranking it up to 1333/1600.
 

Amiga500

Distinguished
Jul 3, 2007
631
0
18,980
0


Depends what your doing with it of course.


If your running programs that use a load of memory (more of a workstation load really), then the AMD will be better, alot better, at least, that is my experience.
 


I don't agree. You haven't supported your conclusions with facts. You need objective data. What if I told you I thought Core 2 Quad was smoother? You wouldn't be able to fight that conclusion because it's completely opinion.

I don't see how Intel can beat Phenom in every benchmark but not be "smoother" than it.
 

Amiga500

Distinguished
Jul 3, 2007
631
0
18,980
0



Sorry - thats incorrect.


Even the old K8 is quicker than the Conroe core-for-core on some of the stuff I'm running.

http://www.anandtech.com/IT/showdoc.aspx?i=3162&p=11




Several of the HPC benchmarks are too expensive for us to test, but we can get some information from AMD's and Intel's own benchmarking. According to Intel, the new Intel Xeon 5472 (1.89 score) is about 26% faster than the Xeon 5365 (1.5 score) when running the fluent benchmark. According to AMD, the Opteron 2350 is about 10% to 60% faster than a 2.33GHz Xeon E5345. That doesn't give us much comparison data, but at first sight it seems that AMD will be competitive in Fluent even at lower clock speeds (2.5GHz versus 3GHz).


I'm not using FEA, I'm using CFD - but the K8/K10 advantage is even more pronounced.
 

meadowlands

Distinguished
Feb 1, 2006
112
0
18,680
0
I must say, I really am confused as to which I should get. Some say "Well, if you have an AM2 board, grab the Phenom, if not, get the Q6600" While I look at it and say, the Phenom is $100 cheaper, but the decent MoBo's are expensive. So if you get the Q6600, you can get a nice gigabyte board for $89.99, for the Phenom a decent board is well over $100.
 

yomamafor1

Distinguished
Jun 17, 2007
2,462
0
19,790
2


I believe both Quantumsheep and TC were referring to desktop performance. On desktop, Intel's FSB is far from being saturated, by majority of the programs. The bottlenecks usually exist in HD, not the interconnect.

However, if you count HPC and server in, then AMD's HT does have its distinctive and superior advantage, especially in multi-core scaling environment. For desktop though, the extra bandwidth offered by HT has minimal impact on performance.
 

zenmaster

Splendid
Feb 21, 2006
3,867
0
22,790
3
If you are seeing "Hiccups", you likely have some configuration problem with your system. It could be drivers, rogue software, etc....

I'm a heavy mult-tasking user and with weaker CPUs, I don't see CPU releated "Hiccups". Such problems are generally Memory and/or HDD related with the memory issues causing excessive HDD activity.
 

yomamafor1

Distinguished
Jun 17, 2007
2,462
0
19,790
2

That is true, to some degree. AMD's HT bus does not offer faster connection, but wider data transferring. That is, in one clock cycle, HyperTransport can transfer more data than FSB. However, if the size of the data pack does not exceed the bandwidth offered by the FSB, you won't notice any difference between FSB and HT.

This is precisely why AMD's HT doesn't offer any real tangible performance improvement on majority of desktop programs. Most desktop programs do not utilize CPU interconnects to their full potential, or even half of it. As a result, Core 2 is able to walk all over Phenom and X2s due to having higher performance, although having slower interconnect.

So, what is smoothness? while one application may take longer time to render on a Phenom because it is a bit slower, that do not prevent other application to move data freely. If you are rendering with a fastest CPU, it may be possible than the other app may slowdown because their data are limited in speed by lesser buses. It doesnt mean the CPU is not fast, it is simply that the subsytem is not fast enough for the cpu to work at full capacity. I'm not saying that the Core2 Quad is lagging... I'm just trying to explain why some processor are executing some program faster than another and sometime slower that the other. Because there is not only a cpu core in a computer..
This is not necessarily true. Like TC said, there is no benchmark that can confirm or deny this. Although theoretically speaking, Phenom does have a better core-scaling architecture due to the shared L3 cache across four cores, but there is no evidence showing if the interconnect is completely saturated by the data.

To be put at a slower scale, imagine running 4 instance of Winrar. Each of them is taking data on one slow ATA33 HDD, processing it, and writing it on the same drive. And all the core has it own file to work with. Poor HDD... Do you think that the HDD will be able to hold on and feed the necessary data to each core?
This is certainly true. For desktop environment, hard drives are usually the bottleneck when running CPU / RAM intensive applications.

So, to answer your question, yes, the phenom can be smoother when running multiple apps than a Core2, even if it takes more time to process the data internally.
I'm not sure how you arrived at this conclusion. For desktop environment, FSB should not be the bottleneck in the entire system, therefore the difference between HT and FSB should be trivial.
 

pat

Splendid
Jan 2, 2001
6,399
0
25,810
19


No, you can have a Biostar T-force 770 for around 82$ US. unless you are talking about crossfire or quad crossfire, then these motherboard are much more expensive.
 

pat

Splendid
Jan 2, 2001
6,399
0
25,810
19


I'm not saying that the Intel is not fast or smooth, I'm only saying that a fast chip is not necessary smoth.If you are in the passenger seat (McLaren has one for demo purpose) of a F1 car, running full speed alorg a race track, and you tri to drink your coffe in a cup withou lid, You'll find your brakefast not that smooth, but fast.

If you are in the pace car, doing the same thing, then the breakfast will not be as fast, but drinking your coffe will be much smooter.


Smoothness is not something that can be mesured easily, but never a faster CPU will be always smooth. But since these day a cpu is very fast, the operation may seem smooth, because human it too slow for them.

As for backing up my fact, if I feven find them, I'l post back.. but having studying microprocessor system at school, programmed assembly, then I read mostly everything about CPU stuff, but don't keep a copy of all benchmark. Too bad I quit programming, because I could have played a bit with them to show you some numbers..

my point here was not to tell which one is faster, but as the OP asked, why a slower Phenom may be smoother than the Intel. Because we all know that Intel is a lot faster than AMD. But smoothness, understanding the inner of a CPU helps when comparing such thing has smoothness..
 

thefumigator

Distinguished
Jul 3, 2005
142
0
18,680
0
-I think smoothness: how quick is your system when a heavy program crashed. Is it able to close a program easily? Ok, the OS has a lot to do here. How does your system responds when 1,2,3 or 4 cores are in full throttle. We are not talking about the app loading times, because that depends too much on the chipset and harddrive speed, and why not, memory bandwidth a little bit.

my scientific aproach of smoothness would be multicore efficiency. On a 4 core system, load a single core application on core 1, while core 2,3,4 are not loaded. Then I would check the performance impact when core 2, 3 and 4 are loaded to 100%. The system that has less impact wins by design. This value is independent on the speed of the core, however, I don't think this is linear, maybe efficiency could get better with different clock speed. Also, in the case of phenom, 4 memory modules may help when you active the option on bios to let each core use a module alone. DDR1066 may have an slight impact too.

the OS is something you have to take into account. But as everybody uses windows, I would make it the standar there.

-HT is a great thing. Despite its not about performance only, but its a better way of interconnecting, Its like a network. It made possible for asrock to mix an Nforce 3 and Uli 1697 on a motherboard, but there are many more examples I could mention.

 

quantumsheep

Distinguished
Dec 10, 2005
2,341
0
19,790
1
On the subject of those server benchmarks in which HTT offered some major advantages i wonder what's going to happen to AMD when Intel release Nehalem. As Nehalem will have an integrated memory controller very similar to HTT AMD's lead in those benchmarks should disappear whici, i believe, could leave them in a fair bit of trouble!

That's all just speculation however.
 

Similar threads


ASK THE COMMUNITY