Intel --- Important ---

[Whizzard9992]

Also I think you need to be clear with your intentions and this benchmark. If you're saying, "AMD performs better at radial blurs in PS," then I don't think anyone will object (or even care for that matter).

If you're saying that this benchmark shows AMD will run photoshop better than Intel, and that we've been mislead by the media, then I think you should spend some more time educating yourself with computer hardware and software benchmarks.

 

[clairvoyant129]

I call the topic creator's benchmarks FUD. There are plenty of reputable sites that CLEARLY shows the advantage of Core 2s.

All reputable sites showing the X6800 ripping out the FX-62 not to mention the 4800+. :roll: Credit goes to gOJDO.

WOW!

:trophy:

Congratualtions, you have posted the greatest BS of the month!

First of all, you said that you have two PCs and latter you are pulling out of your ass, 14 different configurations and numbers. There are hundreds, if not thousands, of reputable sites that have benchmarked a lot of K8 and C2 configurations and came to the conclusion: Core2 Duo at lower freqfency outperforms K8
for all versions of Photoshop, as well as for 95% of the software available on planet earth. For example:

http://www23.tomshardware.com/cpu.html?modelx=33&model1=430&model2=476&chart=186

http://www.bit-tech.net/hardware/2006/07/14/intel_core_2_duo_processors/7.html

http://www.ixbt.com/cpu/images/intel-core2-duo-e6600/results.xls
http://www.digit-life.com/articles2/cpu/intel-core2-duo-e6600.html

http://www.clubic.com/article-36354-9-le-pentium-laisse-la-place-intel-core-2-duo.html

http://www.chilehardware.com/review_200607137.html

http://www.xbitlabs.com/articles/cpu/display/core2duo-e6300_13.html

http://techreport.com/reviews/2006q3/core2/index.x?pg=10

Why do you expect that anyone with brain will take your 100% pure crap seriosly?!? :roll:

For Photoshop CS.2, the E6300 outperforms the 4800+, while the E6400 outperforms the FX-62. There is no match for the performance of the XE6800. Even the Pentium4 and Pentium EE for 50% of the tasks in Photoshop are perfroming better than the K8.

I am waiting for Opteron 175(A8N 5X 2GB DDR-500). After few days I will make your benchmark on my Opteron and on E6600(Asus P5B 2GB DDR2-667) and I will post the results on this thread if it is still alive.

My favorite quote of the day, "Even the Pentium4 and Pentium EE for 50% of the tasks in Photoshop are perfroming better than the K8."

 

[dasickninja]

The initial benchmark would be something I'd call a marginal anomaly. That one time that the one test in that one condition under that specific area that sample b exceeds the score of sample a. I could do the same to any test. Its just can you replicate it on a large scale.

 

[NovemberWind]

Let me start with the difference of the 4600 +10% and the 4800 +8%.

On the ASUS A8N, you can OC with % increase. The main difference between the two is the L2 cache, only that. The 4600 with 512KB x 2 and the 4800 with 1MB x 2. both run @ 2.4 Ghz. That explains the small difference in performance between the two. I only was able to OC the 4800 by 8%, more than that it would crash. The 4600 could handle a little more, 10%. On the P5B, just change the multiplier and the frequency (example: 10 X 340 Mhz)

About my installation:

All the tested computers were fresh Windows XP 32 installed, with all SPs. On the 4 GB RAM computers, the OS would only recognize 3 GB, but that did not matter, because I took 2 GB and tested with the same results (the 2GB were dual channel). After the fresh Windows install, nothing but the Photoshop was installed (of course of the Adobe SPs). I did all that myself, so there is very little margin for discrepancies. The BIOS were all checked for the best performance. I rechecked when OC, and constantly monitored the temperature on them.

Why do I use this method for benchmarking?

I’ve been using this, since Photoshop 6, and as many of you suggested, this is a common and known test in the Photoshop forums. This is not my invention.

My final point is: are the tests being published correct? If this specific test gives the edge to the AMD, does anyone tried any other test that the Intel was not as good? Please, I don’t deserve to be banned for my questions and affirmations, I just want some help, as I mentioned before. This is a real question, not some foolish joke. I have a job, and would not waist my time and yours too. Maybe someone have a coherent explanation.

I am willing to answer all your questions about this matter. Maybe this could help me, and many other users. If I did not answer as fast as you post, is because I have to work too, sorry for that.

Ok, Danica, I'm going to have to agree with Whizzard here - you're clearly not paying attention to this. I'm giving you this last post, and that's it, before I quit being one of the "helpful" few reading and replying to this post.

First off - published results are fine. It is YOUR tests that are likely the source of error. If you insist that everybody else is wrong and are unwilling to take a second look at your methods, you're just wasting everybody's time. Except for the people that started talking about cool car engines.

Secondly - regarding all of that fancy overclocking BS you just wrote above, it most certainly DOES NOT explain everything. You have similar timings all over the place (implying bottlenecking) and you have two times posted for a 4800 chip that are identical, even though one timing is for an overclocked chip and one is at stock timings. (not to mention that the 4600 and 4800 are identical, including overclocking) I call shenanigans.

Next, you didn't answer my question on the Raptors. It was kind of an important one. What hard drives did you use for your benchmarking tests? Was it WD Raptors in all of them? Or at least, all of them with timings <1minute?

Whizzard and I both are on the same page trying to diagnose your issue and we both think it has something to do with memory transfer. His suggestions are good ones and I suggest you try them out. However, I still think you might be dumping too much into a pagefile and thus delaying the data that your processor can process. I still think you either need to get us some information on the pagefile you're using or do some different benchmarks, or both, but at least get some info on the pagefile up on this forum.

My suggestion is that you repeat the test on just the C2DX6800 after eliminating your pagefile (min and max to zero) and see what happens. My hunch is that Windows made a monster pagefile on your computer when it saw all that RAM (I think pagefile size and RAM amount are linked), but like I've said before, I'm no expert. And I believe Adobe likes to dump lots of things to pagefiles. And if all of this is going on, your processor is basically waiting on that poor Raptor to get it the data, and while that Raptor is the fastest SATA hard drive money can buy at the moment, it is nothing but trash compared to RAM data transfer. Not even remotely close. You might also try tweaking the pagefile at very small amounts (maybe 0, 16, 64, 200 MB) and see if that helps/hurts performance. My money is on your timings dropping like a rock once you zero out your pagefile.

So Danica: Please provide some information on the pagefile size you used for these tests (even telling us whether or not you just went with the default Windows size or if you considered it at all before running your experiment would be helpful) and what hard drives you used in all of your setups. (You said the WD Raptor, but some of those older setups don't look like they'd handle SATA. All I want to know is the HDDs used for the timings <1minute, though.) This should help to shed some light on the issue.

After you post this information, please either run some of the benchmarks that Whizzard has suggested, and/or run the experiment I have suggested by dropping the pagefile. Then maybe we can resolve this. However, please understand that your information hardly proves the rest of the world wrong regarding Photoshop performance on a few AMD vs Intel processors. Considering the bevy of data that's out there contradicting you, odds are very high that the error is yours.

Good luck.

 

[Doughbuy]

Can we go back to thread hijacking please? The op still hasn't grasped the fact that he's using one specific photoshop test and then making sweeping generalities, that's like seeing an african american male wear a doo-rag with bling bling and gold teeth dual weilding glock's gangster style and then making the generalization that all of them do...

You have to take things as a whole. Run different filters, change some settings around. Don't get caught up in one little thing.

 

[danica]

Concerning the swap file, I tried to change it from different sizes, with the same result. I even tried without pagefile, but with the same timings. As I mentioned before, the HD activity led does not light during the test.

I also mentioned in my first post, that I use 2 Western Digital 150 GB (they are SATA), in RAID 0.

I insist in this benchmark, because when working in the C2D, you can “feel” that in Photoshop it is a slower computer, you guys know what I am taking about. I am not just BSing about this test, this is program I use, and when you use only one program day after day, you start to get the sensation when you are running in a faster or slower system.

I did OC the X6800, my Ferrari is running premium gas on a good Autoban. In this point, C2D is far, far better than the AMD. Without difficulty I managed to run it @3650 in stable conditions (more than 24%). And as I mentioned before, could not go any further 8 % no the 4800, kudos to Intel. But with the X6800 in this configuration, the filter ran in 30 sec, and all the tests the C2D stock, were much faster than all AMD.

Can you recommend any other test to measure if it is my computer, or any other thing? If you have any recommendation for benchmarking, can you also share with me your timing?

Thanks again

 

[dasickninja]

Try replicating the Tom's Photoshop test.

 

[NightlySputnik]

Let me think, is my gaming fps gain over A64X2 worth less than 1 specific Photoshop test :roll: :roll: :roll: :roll: .

Nahhhhh! I'll get my E6600 anyway in february.

I'll go check where discussion are worth my time. CIAO!!!

 

[Whizzard9992]

Try SiSoft Sandra. It's a suite of synthetic benchmarks, but if you're looking to identify a specific probem (or to ensure you don't have one), that's the tool you want to use.

Also, I agree with Ninja. Try to duplicate benchmarks that exist already. There are a ton of Sandra results out there for AMD and C2D.

I'm glad to hear you've OC'd your X6800

 

[danica]

Thanks Ninja and Whizzard9992

I will try it.

Best regards

 

[Doughbuy]

Synthetics are generally useless. All about real-world testing.

 

[Whizzard9992]

While they're not great indicators of real-world performance, synthetics are good at pinning down problems (such as bad/slow memory, misconfiguration, etc).

 

[cyborg_ninja-117]

This thread failed pretty fast...

 

[djkrypplephite]

Ladies and Gentlemen, may I present to you, the return of the stupid, the idiotic, and totally ignorant, 9-NM!

I don't really post here anymore, but your post was so exceptionally stupid, I just had to comment it.

 

[Doughbuy]

Eh, probably. Most times you would be able to tell exactly which part of your comp is hanging up or causing problems. I've never really used synthetics to test performance unless it's for bragging rights (i.e. 3DMark and such). But using synthetics to find problems is an interesting approach.

 

[joex444]

To the OP:

For example is not proof.

 

[rick_h]

Hi everyone.

I have a AMD 4800+, ASUS A8N SLi 32, 4 GB Ram (CAS 2,5,3,3,7), XFX 7900 GT, 600W power supply.

Recently I bought a Intel Core 2 Extreme, ASUS P5B WiFi, 4 Patriot 1GB (CAS 4,4,4,12), XFX 7950, 600 W power supply, Same RAID 0 as the AMD with 2 150 MB WD 10000RPM. The X6800 was supposed to be much faster, right? But it is not.
SNIP!
”

Something stupid to ponder at; the Intel processors are 32 bit giving them a maximum address range of 4 gigs, which is all consumed with DRAM. I bet if you took out a stick of your memory, (3gigs) your benchmarks would increase. I work with a high-end, dual-core XEON Dell server and we discovered that when we populated them with 4 gigs of memory that the performance dropped. I suspect that the only benchmark where the current athlon CPU can prevail over the Core 2 is in 64-bit operation (when using XP64 of course).


Rick

 

[mandraque]

like i said before...run the test with double the res...

 

[evilr00t]

I don't use this unless it's just obvious FUD:

Something stupid to ponder at; the Intel processors are 32 bit giving them a maximum address range of 4 gigs, which is all consumed with DRAM. I bet if you took out a stick of your memory, (3gigs) your benchmarks would increase. I work with a high-end, dual-core XEON Dell server and we discovered that when we populated them with 4 gigs of memory that the performance dropped. I suspect that the only benchmark where the current athlon CPU can prevail over the Core 2 is in 64-bit operation (when using XP64 of course).

O RLY? I wonder what EM64T is.
PSE/PAE mode.
In 32 bit systems with 4GB of RAM, not all of the RAM is accessible. On several boxes I know, I see 3.25GB of 4GB.
I bet if you took out a stick of memory, you would be running single channel mode and your benchmarks would drastically decrease
Does it use RDRAM or FB-DIMM? More sticks = more latency = lower performance.
O RLY? There are benchmarks out there that disprove what you said. Ask JumpingJack.

 

[dean7]

:!: :!: :!: This just in :!: :!: :!:

evilr00t just bitch-slapped another unsuspecting soul!

 

[gOJDO]

@Rick_H
1.

the Intel processors are 32 bit

No, you are wrong.
The Intel processors are 4bit, 8bit, 16bit, 32bit and 64bit.
Their first 4bit CPU was 4004, 8bit was 8008, 16bit was 8086, 32bit was 80386, 64bit was Itanium. Their first 64bit x86 CPU was Pentium4. Their next CPUs: PentiumD, Core Duo and Core2 are all 64bit.
2.

giving them a maximum address range of 4 gigs,

No, you are wrong.
Including Penium Pro, all CPUs before Pentium 4 can address 36bits or 64GB. All Intel CPUs with EM64T (Pentium 4, Pentium D, Core 2 Duo/Quad) can address 40bits or 1TB.

which is all consumed with DRAM.

No, you are wrong.

I bet if you took out a stick of your memory, (3gigs) your benchmarks would increase.

No, you are wrong.

I work with a high-end, dual-core XEON Dell server and we discovered that when we populated them with 4 gigs of memory that the performance dropped.

It is possible. Depends on the mainboard and northbridge chipset, which is handling the RAM. The CPU is not the reason for the lower performance in this case.

I suspect that the only benchmark where the current athlon CPU can prevail over the Core 2 is in 64-bit operation (when using XP64 of course).

No.
The only benchmark where the K8 can prevail over the Core2, clock for clock, is the synthetic benchmark ScienceMark. In everything else, be it real world application or synthetic becnhmark, the Core2 is faster than K8.
In Unix, the K8 gives the best performance boost in 64bit mode compared to 32bit. Windows XP x64 and 2003 x64 are showing less performance difference, compared to 32bit Win XP/Win 2003.

 

[dean7]

So, I think what you are trying to say is "No, you are wrong"... am I right? :lol:

 

[gOJDO]

right :lol:

 

[dean7]

Well, it was a tough one to figure out. You were kind of sending mixed signals there. I was like "WTF! Why doesn't this guy just speak his mind!"