PCSX2 review : CPU power DOES matter.

[Articbre]

Well thats odd, with my P4 2.8 and my x800XT in FFX and Grandia 3 I recieve almost no difference in FPS when using AAx4. I usually see anywhere between 25-60 fps with both.

[Eric-1987]

lol Idtwo did you mean to say that your P4 is better then AMD's 64 and X2 and opteron line?

[aszucs]

i was sure that overclocking is linear with the speed [heard that pcsx2 very cpu dependant]
but i thought dual core means a little more.. [some said that almost x2... maybe that 19% is because of your gpu limitations? it can be checked with AA/BF]
And AA, BF slow down only if your [and my too] graphic card is not strong enought compared to the CPU.
I think we should make test from the same save point, where there are no optional polygons.. [random walking peoples]
so we could compare our resoults..
When i will be home i will try to make a little test in my P4C.

[GreenImp]

I like it your post man... It's like a Tom's Hardware reiview with eumlators instead of doom 3 & quake 4 as the benchmarks. =P Man this is making me really itch to upgrade my PC. >.<

[Eric-1987]

Well your PC is severly out of date so yeah I would be itching to upgrade it also

[suxxx]

Wow that's a one big good post. So diffrent from all the "omg my game won't work" crap that's flooding the forum. Good job |Gand|Alf|

[RPGW1ZaRD]

Originally Posted by suxxx

Wow that's a one big good post. So diffrent from all the "omg my game won't work" crap that's flooding the forum. Good job |Gand|Alf|

Well I just felt I had to contribute some way to this amazingly progressing project.

I'm interested in programming myself (other than computer hardware & overclocking ) and but these coding skills are way above my knowledge but I'm gonna start studying programming in university and hopefully understand more in the future.

[Coolsvilleman]

Finally. This should be stickied for all the people asking what kindof computer they need to run games at full speed.

[RPGW1ZaRD]

Originally Posted by Coolsvilleman

Finally. This should be stickied for all the people asking what kindof computer they need to run games at full speed.

Nah it wouldn't be fair as no Intel parts are included (I really wish I could include that as well) and it's not like all games run at full speed even with an AMD FX62 cpu. These games run pretty well but there's still many games with lots of polygons that run a bit slower.

In these 3 games however the speed is more than enough to enjoy fully even speedwise (in case the small bugs gets sorted in FFX for example). I get 45 - 172 fps in FFX, average ~80 fps. Grandia 3 runs 30 - 100 fps or so (~60 fps average), and Phantom Brave runs like 50 - 120 (~80 average) fps.

[Mystiq]

Originally Posted by mugener

Please tell me one thing:

Does piexl shader matter on pcsx2???

My card doesn't supoort ps at all, so I can only use gxsd9, but my cpu is 2.80ghz.

Can I get good speed even without ps???

.

The makers of PCSX2 and the video plugins can answer this better, but there are effects possible in computer graphics that are not possible with pixel shaders. Theoretically these effects can just be left out, but you would be left with a half-assed image and it may or may not look right. I don't believe a higher version of pixel shaders increases speed very much, unless these shaders are very complicated; it may just look better.

[RPGW1ZaRD]

I saw some FFX screenies by Luptatorul in the screenshot section with a Conroe E6300 overclocked to 2.8GHz (that's a 970MHz overclock, not too bad ) so I think I'll try and compare my cpu to it and I can already tell it's faster than mine, I'd prolly need around 3.1 - 3.2GHz on this cpu to get same speed. lol

[refraction]

Originally Posted by Coolsvilleman

Finally. This should be stickied for all the people asking what kindof computer they need to run games at full speed.

I agree BUT!!

i think we need more cpu comparisons there, rather than the same CPU set to different speeds. remember the overclocked opteron will have more memory bandwidth than say a standard 2.8Ghz dual opteron, so altho it shows overclocked performance, the variety isnt very great.

Id just like to stress i much appreciate the effort |Gand|Alf| has put in, if we can get it more comprehensive, then i may consider making it a sticky

[techjunkie]

im not sure how the opterons compare to the X2s but would my 3800 (2.0) OCed to 2.8 GHz yield similar results?

EDIT: i have a 7600 GT KO and i get drastic frame rate drop in Kingdom Hearts with even AA 2x why is that?

[RPGW1ZaRD]

Originally Posted by refraction

I agree BUT!!

i think we need more cpu comparisons there, rather than the same CPU set to different speeds. remember the overclocked opteron will have more memory bandwidth than say a standard 2.8Ghz dual opteron, so altho it shows overclocked performance, the variety isnt very great.

Id just like to stress i much appreciate the effort |Gand|Alf| has put in, if we can get it more comprehensive, then i may consider making it a sticky

Oh you're still worrying about memory bandwidth? Well let me tell you this, at least my DFI (=best brand for overclockers currently) motherboard has something called RAM dividers, meaning I don't have to run the RAM at same speed as the HTT/FSB as the CPU speed (why would I otherwise be able to run the cpu at 2.8GHz anyways since my ram only is good to go up to ~255MHz and I run the CPU at 9x311 ). The ram I run at divider 150 or 3:4 of the HTT speed meaning 233MHz. Anyways to prove the RAM speeds barely affect PCSX2 AT ALL, well to me it shows no loss unless I go to speeds like below stock ram speeds like 150MHz, here's 2 screenies below attached that I took some mins ago when RAM was set to 200MHz instead of 233MHz used in the 2.8GHz runs.

Now compare the 1st screen attached below that is 2.8GHz with RAM set to 200MHz to this one which has CPU at 2.8GHz and RAM at 233MHz

The 2nd screen attached below that also has 2.8GHz with RAM set to 200MHz to this one with CPU at 2.8GHz and RAM at 233MHz.

Seems like I missed the highest fps rate in that 2nd screenie for the review as I was even able to snap a higher fps rate this time with 33MHz SLOWER RAM speed. It was jumping between 78.8 but also went to 85.6 FPS for some millisecs.

I wouldn't even have made this review if RAM speeds had affected the results you know. So I think it's safe to say the results of the different speeds reflect the stock speeds of the meantioned CPUs.

At least according to my findings both RAM and Graphic card speed affects the PCSX2 FPS rate nada, nothing. So the results actually are based only on the CPU speed, why would the results be pretty much linear otherwise then if also RAM speed affected it? That would make those bars have an even bigger and bigger gap from 1.8GHz to 2.2GHz to 2.8GHz. But they are linear results as it should be with a CPU dependant software. (take note it should have been 2.3GHz and not 2.2GHz in order to make it perfect linear, 1800 + 500 = 2300 + 500 = 2800 rather than 1800 + 400 = 2200 + 600 = 2800MHz).

[refraction]

ram speed will make a difference, but id rather go for accurate tests. they can be considered unfair is one is getting a non-stock advantage over another

[RPGW1ZaRD]

Originally Posted by refraction

ram speed will make a difference, but id rather go for accurate tests. they can be considered unfair is one is getting a non-stock advantage over another

But what if I run the RAM at same speed tru the whole test? lol

I can run the RAM at 200MHz for all those speeds if I want. But I decided not to do that this time as it never affected the fps rate at least with the CURRENT version.

If you don't believe me I'll just take new screenies of all those places at all the different speeds with ram set to 200MHz and it will still show the same results. (not even differ 0.1 fps unless a program or something ran in the background is slowing me down)

Oh well I might do it anyways so no1 can say the test is unfair even if PCSX2 currently doesn't seem to be affected by ram speed. Then it would be even more accurate in case I missed some highest fps rate numbers that showed up in an certain area.

But for now I'm off to bed.

[aszucs]

Some test result with P4C 2.4@3.0ghz 6600gt(agp, not overclolcked)
512mb ddr ram synced with CPU and FFX (SLUS_203.12)
Tried to test at save points where the scene doesn't change a lot,
so if others test from the same points then we could compare our results...
(memcard, and my pc, and pcsx2 settings attached)

i just put a save state at load menu to be able to load much easyer
and waited a few second after loading, and chechked the fps for 10-20 sec for min/max numbers...
FPS results:
save normal AAx2 AAx4 AAx4+BF
menu: 51.1-51.3 51.1-51.3 51.1-51.3
1: 19.3-21.3 18.3-21.3 11.6-19.0
2: 36.5-39.4 35.2-39.4 23.x-39.x
3: 17.1-19.0 16.0-18.3 11.1-15.1 6.4-11.7
5: 19.7-22.3 18.3-21.3 11.x-17.1
7: 16.5-17.7 16.5-17.7 16.0-17.1 12.8-15.1
12: 42.7-44.6 40.9-44.6 39.3-42.7
13: 26.9-30.2 25.6-28.5 24.3-28.5
I think from cpu point of view only the first column is important,
but i tested with AA too, this way we know wich results are gpu limited,
and wich are cpu [+ram+.. ] limited...
so looks like savepoint 7 is the best (from these 8) to test the cpu...

[Luptatorul]

offtopic

Originally Posted by |Gand|Alf|

Oh you're still worrying about memory bandwidth? Well let me tell you this, at least my DFI (=best brand for overclockers currently)

Unfortunately that's no the case for Intel Core2duo cpus , because on the only DFI board 975X Infinity available for now and tested with Conroe , overclocking is much lower then on my asus p5wdh, and my board sucks, only 425 fsb max stable on current cpu ,waiting for better mobos or bioses for them .

PS:the greatest boards for occ on AMD sk939 platform by far ,if you don't get compatibility issues with the memory, i've had an dfi lanparty sli-d with athlon 64 3700+ @3025 mhz

ontopic

so it is settled
choose some games to be tested :
post configuration on the emu and also cpuz maybe ;
we should see what the requirements of the emu are for now by testing this selected games on different PC's and comparing them , ofcourse pcsx2 would get faster and better as it evolves, but anyway we can get a big picture for now of what it needs

[RPGW1ZaRD]

aszucs> nice little test there, AA4x and BF shows similiar performance loss for me too.

Luptatorul> yea I was only talking about S939 boards but DFI will release new SLI & crossfire lanparty boards September - Oktober for both AMD and Intel which prolly also are quite nice but for many ppl the release date is too late.

I think I'll redo these screenshots once more where I run the RAM speed all the time at 200MHz. But I don't expect much changes.

EDIT: this time around I even try to snap screenshots with EXACT same polygon count and I will also include AA4x + BF shots for the fastest speed 2.8GHz dual core mode. RAM speed is at 200MHz for all games and all cpu speeds. More exact comparisions than this, can't be done.