9600M GT + 9400M vs. 8600M GT!

[Coldwater]

Well I see your point, but in terms of raw power

9600M GT = 120 GFlops
PS 3 = 6 x 204 GFlops per cell processor = 1224 Gflops. (Total system 2 TFlops)

So about 10x raw power.

What do you think? Am I comparing Apples and pears?

Sources: Wikipedia
http://en.wikipedia.org/wiki/GeForce_9_Series#Technical_Summary
http://en.wikipedia.org/wiki/PlayStation_3_hardware

It's all about the code, baby. You'll only see 10x the benefit on something like Folding@Home, not on games where top PC hardware can out-muscle it no problem.

 

[MacBookProJoe]

Sadly, if you want a high end graphics laptop youre going to have to get a PC. I dont understand why Apple wont address this?

 

[appletron]

It's all about the code, baby. You'll only see 10x the benefit on something like Folding@Home, not on games where top PC hardware can out-muscle it no problem.

What about something like video compression? Can the machine use the GPU for that (perhaps with Snow Leopard)? You can't compress video with a PS 3...

 

[Wolfpup]

Well I see your point, but in terms of raw power

9600M GT = 120 GFlops
PS 3 = 6 x 204 GFlops per cell processor = 1224 Gflops. (Total system 2 TFlops)

So about 10x raw power.

What do you think? Am I comparing Apples and pears?

Probably Who knows what the hey they're comparing, because the processors on that GPU are even less flexible than the SPEs on the PS3, but should have more raw power for what they're good at. Definitely higher end parts can bump out more Folding @ Home work than the PS3 (although it's been really helpful to the project too!)

Sadly, if you want a high end graphics laptop youre going to have to get a PC. I dont understand why Apple wont address this?

For one thing, Apple's been obsessed with how their hardware looks for years now. They'd have to make it bigger. I can't believe the number of people who care about that. I mean I love how their stuff looks, but...

 

[Wolfpup]

What about something like video compression? Can the machine use the GPU for that (perhaps with Snow Leopard)? You can't compress video with a PS 3...

Right now the whole area of using the GPU to do non-graphics work is just in it's infancy...but it is coming along.

On the Windows side, we've got at least two programs that are in development that are supporting GPUs for that, with a substantial boost in speed (particularly if you don't have a quad core!)

And Apple is actively pursuing it, it seems. Snow Leopard is supposed to be really pushing that aggressively, so if so, having the 8600/9600 in there instead of that integrated chip will be a big benefit beyond just graphics.

 

[Erasmus]

I'd like to point out that the memory clock of the 8600M and 9600M is 700 and 800 respectively.

Depending if you consider the GDDR frequency doubling or not. But that's irrelevant.

Also, a similary clocked 8600M/8700M/9500M/9600M/9650M/9700M GT should perform almost the same. (<5% difference at max)

Says you, and Wolfpup. Actual reviews disagree with you both.

We'll all know once Macworld or C-Net do a review. I assume you trust those sites to perform proper "scientific" reviews and comparisons?

 

[Wolfpup]

Says you, and Wolfpup. Actual reviews disagree with you both.

We'll all know once Macworld or C-Net do a review. I assume you trust those sites to perform proper "scientific" reviews and comparisons?

Uh....no. Scientficly conducted reviews do not show magical difference between those parts because they aren't there. An 8600 is an 8600-the differences are just in clock speeds on the core and RAM. Cnet does not do proper reviews for the most part. They're basically "oh look, it's blue!!!" Anything that assigns undue importance to Donut Mark is not conducting a proper test. Any wild differences you see in near-identical parts are caused by improperly conducted tests. I could make an identical part seem 50% or more different if I wanted to. It wouldn't mean anything.

The so-called "8700GT" was faster than the 8600GT (it's basically the mobile equivalent of the 8600GTS), and the 9600GT is somewhere in there...but they're all very close because they're all really the same part, just with slightly different clock speeds.

 

[Shivetya]

Again, WRONG. They're basically THE EXACT SAME CHIP. Anyone claiming otherwise doesn't have a clue what they're talking about. It's just a die shrunk 8600GT.

Well according to most charts I can find, using the stand alone part the 8600GT is blown away by the 9600GT. In fact, the 9600GT can beat 8600GT SLI'd


of course the desktop variant is much better than what they shove into notebooks

 

[soLoredd]

It seems all nVidia did was supply 1 more watt of power to the 8600 and up the clock speed from 475 to 500. And by using a smaller die (less heat dispersion) they were able to keep the 500mhz stable. So, they upped the name from the 8600 to 9600 and said "Here you go. New discrete graphics processing."

But, hey, if that guy is still looking to get rid of his 02/08 MBP, let me know.

 

[Wolfpup]

Well according to most charts I can find, using the stand alone part the 8600GT is blown away by the 9600GT. In fact, the 9600GT can beat 8600GT SLI'd


of course the desktop variant is much better than what they shove into notebooks

Exactly, the mobile 9600 is an 8600. The desktop, real 9600 destroys it. The mobile 8800GTS is the closest equivalent to the real 9600GT, and even it isn't as good.

 

[Eddyisgreat]

It seems all nVidia did was supply 1 more watt of power to the 8600 and up the clock speed from 475 to 500. And by using a smaller die (less heat dispersion) they were able to keep the 500mhz stable. So, they upped the name from the 8600 to 9600 and said "Here you go. New discrete graphics processing."

But, hey, if that guy is still looking to get rid of his 02/08 MBP, let me know.

Apparently so. But according to Wikipedia's literature on the 8600M Gt and this independent review, the 9600 has a number of additional transitors (keep in mind, no additional pipelines, and the memory bus is still capped at a mind numbing 128-bit)

 

[therapistfinder]

It appears that the general consensus is that the 9600m GT is a die shrink of the 8600m GT with more transistors and clocked higher. From the 3dmark06 scores that I've seen so far, the new MBP with the 9600m GT in XP typically scores around 5800 or so at 1280x800 (haven't found any scores at default resolution yet). I couldn't find any scores for the previous gen MBP at 1280x800 without overclock but from what I've seen, overclocked 8600m GTs in MBPs produced results in the low 5000s at 1280x800 so I would guess that without overclock, the scores would be somewhere in the mid-high 4ks.

For argument's sake, lets assume that the average previous gen MBP scores around 4800 in 3dmark06 at 1280x800, which I think is a reasonable assumption based on the information that I have. If this is true, then the new one would score around 1k higher, meaning that there would be a performance increase of approximately 20% (this is a pretty rough estimate, but I don't believe that the true value would vary more than 5% from this). I don't know about you guys but in my opinion, that's a decent step up compared to the previous gen MBP.

Now, if you compare this with PCs and factor in the price, that's a whole different story. A lot of people (myself included) balk at the prospect of paying so much for a computer with a mid range gfx card. I think that a lot of people expected Apple to increase the performance to price ratio and were disappointed when they did not deliver. IMO, that's probably where all of this hostility towards the new gfx card is coming from. However, all things considered, it does appear to be a decent step up compared to the previous gen MBP (in terms of 3dmark06 scores).

By the way, I'll post the links to some of the scores that I obtained if anyone wants them. I don't pretend to be an expert, but just from information obtained by searching around, I think that the 9600m GT does provide a decent performance increase, perhaps more than most people think.

 

[JackSYi]

I've just tested gaming performance on my new 2.4 MBP. On games such as Crysis and UT3, there is a noticeable performance improvement.

 

[Wolfpup]

I've just tested gaming performance on my new 2.4 MBP. On games such as Crysis and UT3, there is a noticeable performance improvement.

You'd have to be running the EXACT SAME DRIVERS, and eliminate other variables as well.

 

[therapistfinder]

I've just tested gaming performance on my new 2.4 MBP. On games such as Crysis and UT3, there is a noticeable performance improvement.

If it's not too much trouble, can you specify the framerate you get in Crysis at whatever settings that you played it in?

 

[Eddyisgreat]

Don't forget the DDR3 Ram and enhanced chipset (i would assume having an nvidia chipset would add some benefit)

 

[therapistfinder]

Wolfpup, do you have a personal vendetta against the 9600m GT or something? Almost every single one of your posts is extremely negative and you have little proof backing up your statements. Yes we all know that its virtually the same part clocked higher. That doesn't mean that there isn't a noticeable performance boost compared to the 8600m GT. I don't know about you, but costs aside, I'd rather have a stable 9600m GT than a faulty overclocked 8600m GT running at the same performance level.

I also see little point in testing the performance of both cards only after removing all variables. Even if this was possible, you're missing the point. The majority of people do not care about the specific differences between just the graphics cards. Most of us are only concerned with the performance boost from the previous gen MBP to the current gen MBP. You might be right that there is no difference in performance between the 8600m GT and the 9600m GT. At this moment in time, there is no concrete answer. But all things considered, there is no denying the fact that the new MBP does perform noticeably better in graphics intensive tasks (games specifically) than the previous gen MBP. IMO, it's borderline OCD to have to remove every single variable (ddr2 vs ddr3, drivers, etc) since the resulting information does not have any practical applications and is not representative of overall graphics performance increases in the real world.

Also, to those people who say that the 9600m GT is just an overclocked 8600m GT, yes you're right. The performance increase is marginal at best. But if you put it that way, why don't we all just overclock our 9600m GTs and then measure the difference? It's only fair to compare them this way. Also, based on the experiences of people who already own the new MBP, the new one doesn't seem to have as much of a heat problem (although to be fair, I think the PCMag review said that it did heat up somewhat, but most people that have it haven't really noticed it). Therefore, I would assume that it would be safer to overclock compared to the previous gen MBP, for reasons that need no explanation.

EDIT: Let me just clarify that I'm not supporting Apple's decision to implement the 9600m GT in the MBP. I do think that they could have given us something better, but obviously they have chosen to put their godly profit margins over the needs of the consumer. But based on what I see so far, there is a noticeable boost in graphics performance between the previous gen MBP and the new MBP. It might not as much as people wanted, but I think it's still more than what most people think it is.

 

[Wolfpup]

Wolfpup, do you have a personal vendetta against the 9600m GT or something? Almost every single one of your posts is extremely negative and you have little proof backing up your statements. Yes we all know that its virtually the same part clocked higher. That doesn't mean that there isn't a noticeable performance boost compared to the 8600m GT. I don't know about you, but costs aside, I'd rather have a stable 9600m GT than a faulty overclocked 8600m GT running at the same performance level.

Why on Earth would you think what you just said? Makes no sense. I have an issue with people pretending the 9600 is something it's not. I'd rather have an actually upgraded GPU, not something that wasn't very impressive 18 months ago.

*****I also see little point in testing the performance of both cards only after removing all variables.*********

Removing the system variables might not make sense for what you're talking about, but then you have to say you're comparing the SYSTEMS, not the GPU.

However you *HAVE* to use the same drivers, or the comparison is COMPLETELY worthless.

 

[therapistfinder]

Why on Earth would you think what you just said? Makes no sense. I have an issue with people pretending the 9600 is something it's not. I'd rather have an actually upgraded GPU, not something that wasn't very impressive 18 months ago.

*****I also see little point in testing the performance of both cards only after removing all variables.*********

Removing the system variables might not make sense for what you're talking about, but then you have to say you're comparing the SYSTEMS, not the GPU.

However you *HAVE* to use the same drivers, or the comparison is COMPLETELY worthless.

To tell you the truth, most people do not think that the 9600m GT is that much of an improvement. Ever since there were relatively solid rumors that Apple was going to implement the 9600, many people started complaining that it was not going increase performance by a noticeable amount.

I still do not see the point of just comparing the GPUs. What if there is no difference? Does it really matter? Well, I suppose it's good to know, but what can we do about it? I'm comparing the overall graphics performance of the systems because it is what most people care about.

In retrospect, I do agree that the drivers have to be the same in order to make any comparison. But a direct comparison of just the GPUs and nothing else seems to serve little purpose other than furthering our knowledge.

 

[Hls811]

I also see little point in testing the performance of both cards only after removing all variables. Even if this was possible, you're missing the point. The majority of people do not care about the specific differences between just the graphics cards. Most of us are only concerned with the performance boost from the previous gen MBP to the current gen MBP. You might be right that there is no difference in performance between the 8600m GT and the 9600m GT. At this moment in time, there is no concrete answer. But all things considered, there is no denying the fact that the new MBP does perform noticeably better in graphics intensive tasks (games specifically) than the previous gen MBP. IMO, it's borderline OCD to have to remove every single variable (ddr2 vs ddr3, drivers, etc) since the resulting information does not have any practical applications and is not representative of overall graphics performance increases in the real world.
.

I dont want to get in the middle here, but what have you seen that backs this statement up? "But all things considered, there is no denying the fact that the new MBP does perform noticeably better in graphics intensive tasks (games specifically) than the previous gen MBP."

I'm not disagreeing with you - Just curious because I've been looking for some definitive results and haven't found anything yet... the Primate Labs review posted on the front of MR this morning didn't show anything graphic intensive and from processing power the old and new matched up pretty evenly.

 

[vampyren]

I actually agree with therapistfinder because you need to consider the whole picture and not "just" the GPU thing.

Sure Apple could have added more powerful GPU but would you rather have a laptop with 2h battery life and something you can cook diner on?

I think a laptop should be a laptop and not a beast like Alienware (SLI and shizzle) that weight like 5kg and have a 1h battery time.

Although Apple could have done other improvements to their new laptops like including a build-in 3G or HDMI (which i had hoped for) but any way i am still in love with the new design and slimmer size. I recently sold my MBP and i think this time around i will get me a MB for its size and sexiness

 

[johnsy]

All this stuff doesn't matter. We argue about things of minuscule weight and they worth (in money terms you have to pay) $50-100-200 or more a piece.

There is so much options, but look at them- 2.4 vs 2.5; 256Mb vs 512; regular memory vs low latency, 5200 vs 7200rpm. Might be everything combined would matter more than just adding one of them or just plain stock, but is it worth to spend so much?

for example low latency 4GB memory $78 vs $70 for regular 4GB.
processor and memory upgrade would cost you $200 for older mbp and whooping $500 for new model, OK it would be fair subtract $60-70 for 4 Gb of memory.
For $110 you can upgrade to 7200 rpm HD.

So is minor speed bump worth $388 - 610 ?
Hell at these prices you are very close to mac pro. Of course it would cost you more to add monitor and so on, but it is years ahead for expanding and please, people, don't brag about future, that we will have 10 processors or whatever in one chip- by the time we will have 32Gb of memory will be cheap enough to upgrade mac pro to max.

Now we have completely different issue- portability. Everyone wants to get as much power in as much portable device as possible. But as desktop replacement laptops will lag all the time and in a future even more.

So if someone pays you, well get top of the shelf. In all other cases regular stuff is good enough. Ken Rockwell still uses his 5 year old ibook .

 

[Wolfpup]

I actually agree with therapistfinder because you need to consider the whole picture and not "just" the GPU thing.

Sure Apple could have added more powerful GPU but would you rather have a laptop with 2h battery life and something you can cook diner on?

I think a laptop should be a laptop and not a beast like Alienware (SLI and shizzle) that weight like 5kg and have a 1h battery time.

Although Apple could have done other improvements to their new laptops like including a build-in 3G or HDMI (which i had hoped for) but any way i am still in love with the new design and slimmer size. I recently sold my MBP and i think this time around i will get me a MB for its size and sexiness

For you that's the case, for me it's not. Besides, with a second integrated GPU, have a low power GPU is a moot point anyway.

 

[commander.data]

It seems all nVidia did was supply 1 more watt of power to the 8600 and up the clock speed from 475 to 500. And by using a smaller die (less heat dispersion) they were able to keep the 500mhz stable. So, they upped the name from the 8600 to 9600 and said "Here you go. New discrete graphics processing."

But, hey, if that guy is still looking to get rid of his 02/08 MBP, let me know.

The clock speed of the 9600M GT may only be 500MHz but the shader clocks appear to be uncoupled from the core clocks. Apple slightly underclocked their 8600M GT at 470MHz core and 940MHz shader while the 9600M GT in the new MBP has it's shaders clocked at 1250MHz for a 33% increase. Incidentally, Apple overclocked the 9600M GT's shaders since official nVidia spec is 1200MHz and 1250MHz is the shader clock of the 9650M GS. Memory bandwidth has also increased from 1270MHz (underclocked from 1400MHz official) in the 8600M GT to 1584MHz for a 25% increase. The G96 core isn't earth-shattering, but it would have incorporated the same lossless compression improvements and buffer and cache tweaks to optimize memory bandwidth that have been previously reported present in the G92 and G94 cores over the previous G80 and G84 cores.

I have to admit that I was hoping for a 9700M GT or preferably a Mobility Radeon HD4670. However, a 33% increase in shader clocks with a 25% increase in memory bandwidth isn't insignificant.

It's actually interesting that Apple decided to downclock the core from 600MHz to 500MHz and not peg the shader clocks to 2 times the core, instead devoting thermal room to overclocking the shader clock slightly to 1250MHz since it definitely points to a focus on computing power for GPGPU and OpenCL. They basically clocked the core at 9600M GS level and the shaders at 9650M GS level which I guess they feel cancels out to a 9600M GT.

 

[Wolfpup]

Yeah, those figures aren't bad. So it's a nice little boost, but I still want a truly different part (ie a 9700GTS or better, not an 8600 class part)

Oh well...if the system had Blu Ray, I'd definitely have ordered one by now. I was hopeful based on those rumors

If I got one now, I'd have to buy Windows Vista, and an external drive, and then haul that around, I guess. (And hope that future Blu Ray titles don't somehow stop working on it...hmm if I knew for sure Blu Ray wouldn't stop working, it...) I don't know...I like the design, and it finally has a user replaceable drive, and Apple's warranty is probably better than most company's....