Which CPU for PCSX2 emulator?

Which laptop is this in?

 

It was an Acer 5740 or something. It's 430M+5650+4gb. Seems like a typical base-model specs laptop, which makes the 3820TG or ENVY14 fine for either emulator, I'm thinkin. Cooling is another issue however.

 

@dookie11: Where did you see the 430m youtube videos? I didn't find any. Link?

I'm having difficulties running many PCSX2 games @ fullspeed on my 2.66ghz c2d. The ones I cited (GTA3, Burnout 3, Outrun 2006, NFS Underground, etc.) are just a few. I can run others at fullspeed however, like Kingdom Hearts and SSX. Again I'd be curious to see the results of the 430m or 520m running some of the more challenging games. I've seen youtube videos of the 620m running NFS Underground at 60fps so I have at least some hope.

 

That is true if you are talking about a Core2Quad but not if you are talking about the i7 that has a Turbo Boosted Dual Core mode. So it is a question of what else you want to use your CPU for and cost.

 

@ lackofcheese

Another example I saw was Fight Night Round 3 running at 3.06 Ghz on the mobile i7 620m as seen in CPU z and the Turbo Boost gadget but the mobile i7 820qm was running at 2.0 - 2.13 Ghz with the exact same frame rates. That was the case in some other games as well. The 2.6 Ghz was the highest I saw but it was not in many games at all.

Another example would be Zelda Windwaker in the Dolphin emulator. It hows a Turbo Boost of 2.67 Ghz, Windows 7 Resource Manager shows CPU 0 and CPU 1 at 100% usage, CPU 2 is at about 10% usage (assuming each square represents that amount), CPU 3 is Parked, CPU 4 has around 5% usage, CPU 5 is parked, CPU 6 has a little usage and CPU 7 is parked.

 

I think many are placing too much emphasis on the CPU in relation to how fast a game runs on the PCSX2 emulator. Yes the emulator is CPU intensive and the CPU is important but once you have a reasonable CPU, it all comes down to how well the emulator has been optimised, speed hacks, graphic settings, resolution and many other aspects. You can have a very fast CPU but disable a speed hack and you might still get much lower performance. That could be why you can run others at full speed.

 

I beg to disagree with you. The CPU is *the* only truly important thing in emulation. OpenCL may change that in the future but probably not by much.

We need to keep an apples-apples comparison with each game; I'm running pcsx2 with all recommended speed hacks on, and as I said earlier, games still run the gamut from 20-60fps. When I get a new CPU (again probably the 520m) I'm going to keep the same speedhacks on, so again under those conditions I'd be curious to see how much faster (if at all) games like GTA3 or Outrun 2006 run on that proc.

 

You can connect your Wiimote to your computer via Bluetooth and then tell the Dolphin emulator to use the real hardware as opposed to using the PC mouse to emulate it.

Let me know which other ones you are having difficulty running and I may be able to try them out on my laptop. I just played Burnout 3 - Takedown on the PCSX2 emulator and performance wasn't the real problem. I got anywhere from 35-48 fps but the sound and graphical glitches were annoying. It does not seem to run in some of the builds but a recent svn build ran it. Burnout Revenge is playable and runs at 30fps but the sound completely spoils the experience. I also played Spiderman 2 at 50 - 60 fps but the graphical glitches were terrible, Soul Caliber 2 / 3 40-50 fps and Hulk Ultimate destruction at 40-60 fps but the graphical glitches were bad.

What do you see as full speed? 60 fps?

 

However, the i7-620M didn't necessarily need to run at 3.067GHz to achieve the framerates it did.

The i7 Turbo utility that comes with RealTemp is better at measuring the separate usage of the individual cores. Also, that seems very strange, because I would've thought 0 and 1 were both on the same physical core.

 

here are the recommended specs directly from the pcsx2 website. if you get anything lower then these specs then you will not have the best experience but this also does not mean every single game will work perfect since the emulator is still a work in progress and should be treated as such. also these are desktop part equivalents not laptop so an 8800gt is a desktop 8800gt not 8800m gt.

Recommended Processors:

Intel Core 2 Duo / Core i3 @ 3.2Ghz or faster
Intel Core i5 / i6 / i7 @ 2.66Ghz or faster
AMD Phenom II / Athlon II (X2, X3 or X4) @ 3.4Ghz or faster

Recommended VGA cards:

Nvidia Geforce 8800GT or better
ATI Radeon 4750 or better

Other Hardware Recommendations:

512 mb or above of System Memory:
Windows XP With Service Pack 3
Windows 2000
Windows XP 64 With Service Pack 1

1 GB or above of System Memory:
Windows Vista 32/64
Windows 7 32/64

 

Hmmm. The difference in clock speed requirement between the i3s and the i5s doesn't make any sense.

 

@Shoegazer:

YouTube - PCSX2 Beta 1888 emulating God of War II [NTSC-U] Half Speed on notebook
YouTube - PCSX2 Beta 1888 emulating Kingdom Hearts II [PAL] Full Speed on notebook
YouTube - PCSX2 Beta 1888 emulating Final Fantasy XII [NTSC-J] Full Speed on notebook

I typed in the search bar "pcsx2 ati 5650".

Similarly, here's a video for Dolphin, with my guess is a Wiimote (mouse): YouTube - Dolphin Emulando Resident Evil: The Umbrella Chronicles Wii

EDIT: This guy also plays RE5 on the same comp: http://www.youtube.com/watch?v=furXYqFe_Xo

All these videos emulate/run well using an i5-430m, so I might consider just getting a 450m to save $ and battery life. What do you guys think?

 

Thanks, but this post is of no use. We've all seen this in the forums for pcsx2. What we need is the laptop equivalents, if someone can convert those (e.g. converting-inches-to-centimeters-type-conversion)

 

With CPUs, two Nehalem cores are two Nehalem cores. There are differences in cache and memory speed between some of laptop CPUs vs desktop counterparts, but by far the largest difference is clock speeds.

However, because they give different figures of 3.2GHz for i3s vs 2.66GHz for i5s and i7s, I can't really say which one to concern yourself with.

 

the only thing i could do for what you want is compare the cpus that i currently own as i have done several tests on them and i found that my x9100 running at the 3.06ghz is just slightly faster then my desktop e6600 at 3.2ghz

finding numbers for performance between cpus shouldnt be that hard but yes there is a difference between i3 and i5. cant say if this applies to mobile cpus but i beleive the i3 cpus do not have turbo boost among a few other minor things.

 

Sure, but Turbo Boost changes the clock speed, so having a different clock speed requirement seems to be pointless here. Besides that, dual-core Turbo Boost in most Core i5s is only 133-266MHz, which still doesn't explain the 533MHz difference in their requirements. Apart from Turbo Boost, there's a difference in that the desktop i5s support VT-d, Trusted Execution Technology, and the new AES instructions, but I don't think that PCSX2 would make use of any of these. Consequently, the figures still don't make sense to me.

 

Just tried FFXII on my G73 with the newest PCX2 Beta and it ran fine with everything stock settings. 60fps in all the FMV/Video and like 45+ during the game.

Im sure with tweaking and overclocking things can get even better.

The G73 is using the 720QM and a 5870 card, not sure if the gpu is used really unless they have made advances in hardware acceleration since I last used an emulator. It was all cpu back then.

 

Alas, all those videos *don't* run well at all. GOW2 and RE: UC are at least only half-speed, far from playable, and actually a bit slower than my 2.66ghz c2d, and Kingdom Hearts/FFII actually ran fullspeed on even my older system. So it definitely looks like the 420m is not at all a good proc for emulation, far worse than the 620m which able to run even the most difficult games fullspeed, as this guy has proven:

YouTube - frankyfife's Channel

 

It seems unlikely to me that the difference between the CPUs would be solely responsible for the difference you're suggesting. With full dual-core Turbo, the i5-430M runs at 2.53GHs vs 3.067GHz for the i7-620M, which is a 20% difference in clock speed. Somewhat large, but it's hardly going to make the difference between half-speed and full-speed.

 

That is what I was trying to say earlier. Many are placing too much emphasis on CPU speed and forgetting that it also makes a huge difference about the actual code of the emulator, optimisation etc. Particularly the dynamic re compiler efficiency.

Try playing half of those games that are full speed now on a really early release of the emulator or with the interpreter and they would not be running as fast at all.

God Of War 2 still ran at that same speed on a 3.06 Ghz Dual Core mobile i7 620M which is related to my above point. Also take into consideration that I was running it at a higher resolution of 1600 x 900 as well. I bet there are no games that play better with the i7 620M Vs the i5 when using the same settings, plugins and svn.

Having said that, I have seen God Of War 2 running on PCSX2 at full speed but those were mostly Desktop i7's OC to 4 Ghz. So maybe there is a difference but only with a significant jump in clock rates as opposed to just 15-20%.

For example the following Desktop specs appear to run God Of War 2 at full speed:

 

And again: I am comparing *apples to apples* by standardizing on the PCSX2 version used, speed hacks used, and even the game in question. For example, I have noticed that with frankyfife's test results, using the same PCSX2 version and configuration, in NFS Hot Pursuit 2, he was able to get full-speed on his 620m. Contrast with my 2.66ghz c2d platform, which has more cache memory than his has, and running at PS2-native resolution (lowest setting), I'm lucky to get 35fps on the same game. Check here:

YouTube - frankyfife's Channel

That has *everything* to do with the CPU and not much to do with anything else. Granted, it may be related to the architecture (i7 vs. c2d) as much as raw speed, but my point is the same either way.

I doubt that, based on what's already been posted here on the 420m and what frankyfife has demonstrated with his 620m. Still, I'd be interested in seeing PCSX2 results from a 520m platform, which was what I was hoping someone would provide in my original post since I was considering picking one up.

 

The 520M is only 133MHz faster than the 430M (i.e. 5%), so I find it unlikely that there would be much of a difference.

I also think the testing conditions aren't strict enough for the 430M vs 620M comparison. I do suspect that the 620M would perform noticeably better than the i5-430M, and the purchase could be justified if running PCSX2 is very important to you, but I don't see how the 20% difference in clock speeds could cause more than a 20% difference in framerates.

 

Does this video not show the game running at a good/playable speed? YouTube - PCSX2 Beta 1888 emulating Final Fantasy XII [NTSC-J] Full Speed on notebook

If you fastforward to about 7:23, you'll see that he's fighting the AI at a reasonable speed. Seems fine to me.

Also, how does this CPU compare to i5-430M/520M ? YouTube - [Dolphin SVN 2786] Tales of Symphonia (Perfectly!) - Part 1/3

And if this is a legit video, according to the specs, it seems fine enough: YouTube - Dolphin SVN 2939 Super Mario Double Dash for GameCube

This one, if legit, is using a i5-520M YouTube - NFS ProStreet max settings

Seeing that last video, makes me feel that a 520M's purchase is warranted if I want to fully enjoy PCSX2 or Dolphin (both NGC and Wii games).

 

Does this video not show the game running at a good/playable speed? YouTube - PCSX2 Beta 1888 emulating Final Fantasy XII [NTSC-J] Full Speed on notebook

If you fastforward to about 7:23, you'll see that he's fighting the AI at a reasonable speed. Seems fine to me.

Also, how does this CPU compare to i5-430M/520M ? YouTube - [Dolphin SVN 2786] Tales of Symphonia (Perfectly!) - Part 1/3

And if this is a legit video, according to the specs, it seems fine enough: YouTube - Dolphin SVN 2939 Super Mario Double Dash for GameCube

This one, if legit, is using a i5-520M YouTube - NFS ProStreet max settings

Seeing that last video, makes me feel that a 520M's purchase is warranted if I want to fully enjoy PCSX2 or Dolphin (both NGC and Wii games). But...the original videos I found were using the 430M, so I'm still stuck btw getting 450M vs 520M on the ENVY.

 

Unless PCSX2 makes use of AES-NI or VT-d, performance will be the exact same between the 450M and the 520M, because their clock speeds will be the same.

 

ok im posting a few cpu comparisons ive done among friends to try and give a performance comparison with the pcsx 2 emulator.

firstly
my x9100 at 3.06ghz is slightly faster then e6600 at 3.2 ghz

i5 520m is a bit faster then t9500 with turbo boost on

i5 540m is faster then t9800 but sometimes faster then t9900 (by a hair) on some games

with that said i think that would make the i5 450m and i5 520m a little bit slow for pcsx2 but will work with most games.

but like said above the i5 520m and i5 450m should be the exact same in terms of performance for pcsx2. the most important thing to remember about the i series is that turbo boost will only work if tdp allows it so if it your laptop model your looking at has a history of being hot then you might not even get the benefit of those higher clocks.

 

Wow, even for a low-tecchy person like myself, I even understood this complicated/tricky comparison thing haha. So 450M/520M no difference basically besides max turbo boost? Looking at the comparison on intel's site, I notice that there's "4.8 GT/s" for the 450M vs. "2.5 GT/s" for the 520M regarding "System Bus." What exactly does that mean, and which of those is better?

Also, since they're basically the same in terms of base clock speed... Another question: Are i5-xxxm CPUs OC'able like desktop CPUs?

 

I guess I'm not making this clear enough. *All* of those above games work fullspeed on my current 2.66ghz c2d, without any problems at all, excepting NFS ProStreet which isn't even an emulated game (he must be running the PC version, which I could care less about). But I don't care about any of those games since I already know they're fine. My goal is to find out which games I found *challenging* on my current c2d will work well on a 520m. That includes NFS Underground, Outrun 2006, GTA3, etc. Otherwise, no need for me to buy a new notebook. Capiche?

 

Wasn't this my thread?

Anyhoo, what I'm trying to find out from all these posts is whether or not the 450M will be suitable vs. having to spend $100 more on the 520M.

 

Sorry, didn't intend to hijack you there. I'd be curious about that too. I'm more partial to the 520m but either one would be interesting - and again, for challenging games only (which is what you'd want to see mostly as well - right?).

 

The i5-450M and i5-520M both have a base clock speed of 2.4Ghz and Turbo Boost to 2.67Ghz on two cores, although on one core it's 2.67GHz vs 2.93GHz - since PCSX2 uses two cores, I wouldn't expect this to matter much.

Apart from that, the i5-520M has three features that the i5-450M doesn't - VT-d, the new AES instructions, and Intel Trusted Execution technology. That last one is a security feature, while I doubt PCSX2 uses VT-d or AES-NI. As such, I doubt the i5-520M would be any better for PCSX2 than an i5-450M.



As for i5-520M vs 2.66GHz Core 2 Duo, the i5-520M performs something like a 2.8-3GHz Core 2 Duo in most other tasks. If you want a big improvement, you need an i7-620M or overclocking.

 

i dont think you understand the concept of emulation. there are no challenging games. the only thing challenging is emulating particular parts of the console itself. when bugs are present it will be for all cpus even if you had a super computer from the future it will still be there regardless of your cpus speed.

the recommended isnt really that high anymore. as for the 520m its been stated several times when it comes to the emulation of pcsx2 it will be the same as a 450m providing your laptop choice has the thermal abilities to allow for the maximum turbo boost.

 

Ya, in many ways I agree with ya

LaptopNut tells me that the 450M should be fine as a CPU. He also told me that these emulators though they're CPU dependent, it also depends on your settings, speed hacks, etc etc etc. However, I'm still unclear about 450M's abilities. The 430M seemed fine in most of those videos, and the 520M seemed great in its videos... but they're not definite answers to warrant a $1000 purchase quite yet.

 

I tend to agree to an extent. I have seen some unofficial / private builds of PCSX2 that have been optimised only for speed pretty much ignoring compatibility (in some cases making it less). These builds are frowned upon on the official PCSX2 forums by the way so I wouldn't mention them there.

In this case, people who tested the emulator out with the same CPU were seeing an extra 30 fps in some games. There was a Japanese build that played Shadow of the Colossus at 60 fps where the official build would play it at 30-40 fps for that particular person with the same configuration.

I will try to get hold of one of these builds for some testing.

Edit: Another example.

In God Of War 2 using the latest official build of PCSX2, I get around 30-35 fps with all recommended speed fixes enabled and at my native resolution of 1600 x 900. If I enable all speed hacks I get a few extra fps but not much.

I just tested an unofficial build called pcsx2 maetel 2.6 and when I played God Of War 2, my fps was a smooth 50-60 fps never dropping below 50 at the same resolution and the graphics quality didn't look any different. I will try to get a more up to date build because this one was compiled in 2009 and is not using the latest GUI.

Some one talking about the unofficial builds speed...

Edit: Further testing suggests that pcsx2 maetel 2.6 is specifically optimised for the God Of War series only.

I tested out the first God Of War and I get 60 - 75 fps even when there are lots of enemies during battle.


Edit: I have found more Pcsx2 Maetal builds for other specific games such as Tekken 5 and shadow of the Colossus. If you try to use the builds with other games, they are extremely slow.

 

That sounds very interesting - why are they frowned apon?

 

hi , i'm trying to make the emulator to work but i'm getting problems.. how do u set it up?

 

I think the reason is due to people on the forum asking for support and questioning the developers about why an unofficial build runs faster than their official one. The users easily forget the time and effort required to balance speed and compatibility but the unofficial builds are often optimised or hacked for specific titles so they don't need to worry about anything else.

 

I understand perfectly well the concept of emulation but thanks. There are of course challenging games; games that run sub-optimally (read: with fewer than 60fps on an NTSC game) on modern x86 or x64 platforms are "challenging" to the platform no matter what is going on under the covers. By this definition, GTA3 is challenging to emulate at playable speeds, and Kingdom Hearts isn't. It's really that simple. Games that are challenging for an emudev to emulate, due to missing opcodes, graphics/sound subsystems, protected code, etc. is a different subject altogether.

And again, I'm not interested in comparisons between the 450m and the 520m; more between the 520m and a 2.66ghz c2d (or if you prefer, the Intel T9550 CPU).

In any event, the issue at hand is whether the Core i5 is worth the money as an emulation platform, and as has been discussed and revealed here, neither the 450m nor the 520m are wise investments at all when considered purely for emulation, especially considering the c2d gives you the same amount of power at much less cost. So I think I've got my answer and will wait for the much higher-clocked mobile CPUs to hit the market in Q4 2010.

 

That is more likely due to the configuration of the official build. Try, for example, switching from MicroVU (which is default on later builds) to the SuperVU recompiler, and like magic, the official build is just as fast as that hacky stuff. Generally speaking it's wise to stay far away from hacks like that as they tend to be riddled with viruses, strongly discouraged by the devs, and are otherwise just a waste of time. Think about it. If those people really knew what they were doing they'd be contributing to the official project.

 

ummm...

I'm gonna just go with piracy isn't supported on the boards.

rent/buy the games or if you really wanna find stuff well...google searching can help if you know what you're doing.

 

You make some very good points. PS2 emulation is still very new to me, I only started messing with it after becoming curious about how far along it had progressed after seeing the OP's question. I did a few tests and yes, it does seem it is more down to the configuration, particularly the MicroVU Vs the SuperVU recompiler.

 

Glad to help. PCSX2 has come quite a long way in a short period of time, and is bound to get better in the near future with some interesting developments on the horizon. If you haven't already, you may want to check out Dolphin, which is progressing even more rapidly, and nullDC since it has recently gone fully open-source like the others. And if you are at all inclined towards "old-school" systems, you can't beat MESS for the breadth of systems it covers, many of which it emulates quite well.

 

I have been using Dolphin for a short time with games such as Zelda, Mario, Fire Emblem, Buffy, Metroid etc. I am shocked at how well it emulates and at how good compatibility and speed is. Many games have been completed right through to the end by some. I haven't tried nullDC yet though. I will look into that.

Another great thing is how cheap all of the games are. I purchased some used copies for as little as $5 from Ebay. I had never heard of MESS before, now that looks like an amasing project to be emulating so many different oldskool systems.

 

In 06/07, I tried NullDC and hated it with a passion. Before I even went to college in 05, I had been using Chankast for several years, which IMO is still the best emulator for DC.

Since PCSX2 and Dolphin are very demanding, I'm wondering whether or not I should even get a laptop soon, or just wait till Q3-Q4 for better CPUs.

I sure wish someone who actually had an i5 or 620M would chime in...

 

shoegazer ive given a few comparisons to the cpus you are asking about already all you gotta do it put that info together.

im sry but i just cant accept that a game running on an emuator is in anyway challenging to run as its not the games fault here none of the code built into the game is causing this sow down. its all down to the emulator which my point was proven by using hacked builds.

let me elaborate here for a bit. think of the example you gave (gta3) you say its a challenging game to run ok fine but from a dev point of view that doesnt help much. since it runs fine on a real ps2 thats not much info to go with there. now say you told the dev it runs bad because of the microvu or something to that effect and find another game that has the same issue. from here the dev will find out why its having problems with that specific part of the emulation and can fix it thus possibly fixing it for all games that use this particular part of the code not just the two tested. i know my explanation skills suck but i hope you get what i mean from that

i figure hacked builds are frowned upon for all the reasons posted up to this point and adding in that honestly there is no telling what all this added in/ changed code will do to your pc and knowing how some people can be will blame the original devs.

i havent tried nulldc in awhile since its never worked good but i can vouch that chankast is great. although it cant run sword of the berserk right

 

I am beginning to wonder about peoples VU / Cycle Stealing configurations with regards to their performance posted in youtube videos, particularly for the God Of War series. I wonder if people have failed to mention that when they talk about the other speed fixes.

On the latest official build, I gain an extra 20 - 30 fps depending on those settings in God Of War. Some games will not work well with those settings but others will probably be fine. With cycle stealing on default, I get 30 fps, with it at medium I get around 50+ fps.

From the pcsx2 official forums.....

 

NullDC has alwas been, even from its very first release, orders of magnitude more accurate and compatible than Chankast. That's a pretty widely known fact - Google around and see for yourself. Chankast's famously poor accuracy could be explained by the fact that it's littered with hacks, was completely abandoned by its devs many years ago in a very premature state, and everything is HLE'd.

If you want a real competitor to NullDC, try either Demul or Makaron (both of which, along with NullDC, support the Arcade/NAOMI side of the Dreamcast, which Chankast never touched). NullDC is not without its faults either (and there are many) but for pure Dreamcast emulation, nothing beats it overall - especially considering that it's currently the only open-source Dreamcast emulator. Give it a second spin and you'll see what I mean.

 

You completely missed my point. I'm sorry, but I'm just not going to spend time in furtherance of this discussion if you don't get it.

 

@ ryokowashu regarding hacked builds

Bare in mind that the reason why speed hacks work are because they tell the emulator to skip over potentially important tasks that are completed under normal circumstances (which takes more CPU cycles). Due to this, the CPU dealing with the emulation has less to work to do resulting in a speed increase. The speed hacks allow less powerful hardware to play the same games faster due to this but due to the jobs they miss, we sometimes get missing graphics or other abnormal behavior. A much more powerful system with a superior CPU will be able to deal with accurate emulation and have less need for speed hacks.

Both the speed of the CPU and the emulator code are important for the speed of the actual emulation. I think you will only see differences in performance between different CPU's if the clock rate differences are very substantial though.

 

So 450M? Or should I go higher?