MacBook Pro Santa Rosa 128MB vs 256MB - Benchmark compilation

Ever since the new Santa Rosa MacBook Pros came out, a lot of people have been wondering how well the 128MB VRAM MacBook Pro stacks up against the 256MB model. While there is some concrete information in various forums and threads, it is generally hard to find and hence hard to compare. In order to help prospective buyers make a sound decision, I thought it would be nice to have a compilation of what we know so far. Here we go:













Hope this helps, and by all means, feel free to add your own benches!

Edit 3: Updated 3DMark06 and In-Game Benchmarks
Edit 2: Allright, that's fixed. Thanks Nicholie!
Edit 1: owwwww massive mistake in the thread title... is there a way to change that???

 

Nice thread. I've got another score if you like. I Just ran 3DMark06 on my 15" 2.4GHz MacBook Pro (256MB 8600m GT) and got 3672 at 1400x900, everything else on Default. I am running Boot Camp 1.3 Beta and Vista Ultimate (x86). I used 3DMark06 Professional Edition 1.1.0.

 

PM an Admin, Andrew or someone can take care of it.

Interesting scores, thanks for getting that all together. Makes me happy I saved $500 and got the 128mb version. Not a serious gamer here.

 

Good, I'll add it when I have a few more values to put there!

 

There so many questions about the new MBP, let's just merge all those threads into this one to make it easier for someones who's interested in purchasing the MBP.

 

Those are some pretty non-uniform scores. Is it all due to different drivers for the same amount of vram?

 

wow this is good stuff, i am definitely going for the 256 version

 

Great thread, thanks for making this. Should aid in my purchase decision

 

I think a lot of it might have to do with overclocking. Some of the scores for the 128 MB VRAM MBPs in 3DMark 05 vary by 20%.

 

Come on 128MB version, get some higher scores darnit! *roots*

 

Actually, I did not include any overclock scores, so unless some people forgot to mention they were overclocking while they ran the tests, these should all be stock scores. And indeed, I do agree that these scores are really inconsistent. In fact, the third table up there (which shows a relative comparison of the 128MB and 256MB 3DMark06 performance for each 3DMark06 sub-test) comes from data posted on the insidemacgames forum by BareFeats, the same guy who did the famous comparative OS X gaming test; they give a radically different picture than the OS X results he posted on his web page, which showed a negligible difference between the two models. For what I know, he refrained from putting these results on his web page, as he had a really hard time coming up with these numbers. The tests gave a very wide spectrum of scores when repeated in seemingly identical situations.

 

This is upsetting. I was hoping there wouldn't be a huge difference between the 128/256MB 8600GTs.

 

the scores will be more consistent once the drivers are ironed out. Some people are using modified drivers.

 

Hi

When i see your table about 3D mark, i would like to add something that i hope you have already heard, it's about the frequence of GPU between MBP 15" 256VRAM and 17" 256VRAM.

I let you the url of the bench : http://www.barefeats.com/santarosa.html

Therefore it will be better if you add information about screen size 15" or 17" that can show the difference of powerfull.
sorry for my bad english

 

Overclocked my 3DMark 05 score is 7816 at defaults =/

 

guys, let's keep some ingame Benchmarks coming with varying AA AF settings.

FarCry
S.T.A.L.K.E.R.
Oblivion

etc.

I'm sorry I won't join your efforts until I'm sure what to buy...

 

Add Supreme Commander, Battlefield 2 and 2142, and C&C 3 Tiberium Wars to that list

 

new benchmarks:

 

Thanks, I added the Prey results to the charts. I won't add the other two results, since they are sub-tests of 3DMark06 (and they don't really fit in any of the charts).

 

first- nice thread. im happy my benchmarking didnt go to waste.

second- the reason there seems to be a wild variation among scores and performance (even with the same hardware) is because we still have VERY immature drivers. literally released within a week of the hardware, and thats all we got. some people are using modded drivers, others are sticking with the apple stock. we will see better drivers and better performance with time.

also - i would like to add two subjective benchmarks:

Tom Clancy's Rainbow Six: Vegas
it wouldnt run at an acceptable framerate REGARDLESS of the resolution i used. this means that there are driver problems related to the game. going to high resolution from low didnt make the game run any slower. drivers will have to mature for that game to run well. also- i heard that it was horribly unoptimized in the first place. we might need to wait for the next rainbow game to have a good experience. my guess was that it was running at 20-22 fps regardless of the action taking place, screen resolution, "settings" quality, moderate anti aliasing, etc.

and

GRAW 2
basically the same experience as above, except that by drastically lowering the quality i was able to push 32-35 frames per second. again, that framerate was scarily consistent regardless of the action on screen. that means that there are driver issues with this game also.

in a word, we can expect MUCH better performance with both of these games with time as newer drivers are released.

2.4ghz macbook pro. windows xp 32 bit.

 

Just one tid-bit to add in the consideration: The 256MB version has the 2.4Ghz C2D as while the 128MB has a 2.2Ghz. This may also account for these speed differences. Not the VRAM alone.

 

well... that makes up for 9%!

 

Hey, I heard that the Asus laptops' in game performance doubled when there was 3 gbs of ram in the system (since Vista can then use more system ram for gaming). Anyone able to test this with their MBP? Could we get a 2.2 ghz and 2.4 ghz to test this?

Here is the Asus thread... http://forum.notebookreview.com/showthread.php?t=125246&page=38 (FPS pretty much doubled with 3 gb)

 

Sure... just send the cash for the RAM on over to my paypal account and I'll get right on it!

If it helps - when I upgraded the RAM in my desktop from 512MB to 1GB, I up the details in Doom 3 from low to about med-low, and reduced loading times in Half Life 2. So yea, more system RAM should help with something.

 

I think doubling might be a bit optimistic!

Josh

 

I have no experience whatsoever in shared RAM under Windows, never tested the impact. Anyone?

 

Cutting through the benchmarks, can someone just tell me one thing. Will 2.2 Ghz MBP (128 mb one) play new games for 1-2 yrs ? (I am not looking for maximum settings/max resolution.... i just want playable frame rates and decent effects)

 

Yes, it will play games new games fine. Though on some games, you will not be able to have as high of settings if you had a 128 mb compared to the 256 mb.

 

So Macbook Pro 2.2Ghz it is for me

 

I'm buying one in a few hours...

 

unfortunately i have to wait for 2 weeks min to get my hands on one

 

yeah... the shipping times are increasing every day... I wish I had ordered 2 weeks ago right after they were announced! dammit!

 

Well, just to re-emphasize my point about the 128MB version. After overclocking (I did it generously, not stressing it too hard - also with hardly any noticeable heat difference) I was able to get my card running as fast as the 256MB Counterpart at least at their stock speeds. Yea, I know they can overclock too, but it is nice to know the majority of the basic speed difference can be easily made up by those who don't mind a small risk (Almost zero percent chance in my case imho when you know what your doing). This according to the benchies earlier in the thread since I don't have a 256MB with me to do my own comparitive testing, my scores are even faster than some of the benches of the 256MB version.

 

Nice. I guess I'll go with the 128 mb one too then. I really cant justify the higher price for myself.

 

Just to support what I just said, here are a few examples:

top of the list 3dmark 2005: Jacquez - 1024x768 - 6800
Mine was: blull - 1024x768 - 7000 normal, ~7800 overclocked

In 2006
the two bottom 256 scores were in the 3400s, my overclock was at ~3500 at the same resolution 1280x1024

Also, towards the bottom of the list there is a 256MB score for the lost planet demo - both my directx 9 and 10 versions get better scores, my DX 10 score was right at 17 and 23FPS, and my DX9 scores were 20-30 stock and around 27- 36 overclocked.

 

If i ever run into you, your beer is on me dude

 

Don't say that lightly, I might take you up on that! I'm going to try and get some benchmarks up to show off my baby a little bit more. I have a question though, just to make sure I have this right, about the Fear Demo. Am I correct in assuming that the Demo version does not include the fear benchmark? I really wanted to get some scores up for comparisons sake. The demo runs very nice on my machine as far as I can tell, however I don't have fraps up so it might just seem like it's running smooth =)

 

Dont worry mate, a beer is a small price to pay for your benchmark Alas, my old thinkpad wont even dare to try and install FEAR. I am eagerly waiting for my MBP to arrive so that i can buy all those nice homely games (such as FEAR, Quake 4 ) and run them on my baby

 

Hi.PRISM,
you are doing great job, i want give u my new test of 3dmark06
i installed 3DMARK06 complet version for this time
under XP
8600GT 256M

3dmark06
1280X1024: 3760 points
1440X900: 3781points


jacques

 

This is kind of a dumb question but does the shared system memory over to video memory only work in vista? Or does it also occur in XP? If it is Vista only would that not make Vista far better for gaming?

 

I honestly don't think the 256MB version will run future, extremely graphic intensive games at higher then low settings either; given the fact that the card is only an 8600 model. If it were a 8700/8800 model with 512MB, then you'll be able to turn up the detail. I'd definately expect Crysis and UT07 to run, but at low settings no matter how much VRAM you have on an 8600. Less intensive games should be a breeze however.

 

You would be surprised what can be done with your macbook pro mr.pibb. Actually, with turbocache and the way regular ram is slowly becoming a bigger and bigger supplement for vram, I honestly don't think the 128MB ram is going to be the hold back. Processing speed? yea, eventually. I can run F.E.A.R with all high detail and my card runs smooth as pie. Oblivion runs great too.

As far as the 8800 model with 512MB, its only time before that becomes out of date so it really is the same as the 8600. Sure, by the time the 8800 has to run on low settings to run a specific game, the 8600 will probably not work either, but that is how the technology industry goes. But though, if you are comparing the 8800 to the 8600 then it's not like we don't know who the clear winner is. If it worries you that the 8600 isn't as fast as the 8800, then ya might be better off buying a desktop machine to keep up with the times, since it is much easier to upgrade and change parts.

But, if your looking for a good mobile machine, with great battery life and the potential to play current high end games, and potential future high end games, even if they are at a lower setting a year or two from now. At least you will have something that will be able to last and still play games for years to come =)

Hope that makes ya feel better =)

 

blull, what program are you using to overclock?

nTune seem to reset clocks after restart and the other two programs I've tried can only read the 8600M's lower/idle clock steps.

 

It all depends on what you consider 'great.'

I was hoping that the MBP/8600 combo would run Oblivion better than the X2/R1900XT combo in the desktop I had last year when the game came out. Taking Oblivion on the go and keeping a machine that can run OS X is pretty sweet.

From what I see of the scores, the new C2D/8600 combo is about the same as a top of the line desktop from two years ago in absolute terms. That is great, but a bit less than I had hoped for.

 

Overclocking is a dangerous game. It really is not worth it.

In games you will at best get an inperceptible gain in framerates, that will deteriorate once the chips temperature-triggered clock throttling kicks in.

Heat will make your whole machine run slower if you actually play games with your system overclocked, unless you get a cooling pad or something augment the books cooling. Heat from the GPU will bleed to other areas with temp sensors.

In the end you get some short-lived bragging rights for good scores in benchmarks, but an impressive 3dMark05 score these days is 35,000. A notebook can't compete with that so why bother?

 

Nothing that really occurs with this mild overclocking. The temperature of my gpu pretty much didn't go up at all.

 

I don't know what happens if you overclock it too much and then don't see anything anymore because the GPU crashes? Will a reset do the trick or will the card be messed up because you won't be able to downclock it while you can't see?!?!?

 

Yes, for most machines - but that is my point, temperatures that are virtually the same overclocked as they are underclocked will not have those types of problems - enter the macbook pro. Also, overclocking isn't really overclocking that much since the chipset has been underclocked by apple compared to the standard 8600m GT clock rates. My computer is rock stable at the overclocked speeds and it really does not cause me any problems at all. I don't feel my machine crack under pressure or anything of the sort.

Let me re-emphasize, I think the most I saw the temps increase after my high overclock is ~3 degrees C.

 

Does the MBP's 8600GT support TurboCache? It would be nice if it did.

 

Vista x64 shows almost 900MB of usable memory. I assume thats considered turbocache? At least I thought thats all turbocache was, using ram for usable vram.