NVIDIA’s GeForce 600M Parts

We just covered the AMD side of things, but yesterday NVIDIA quietly refreshed their entry-level and midrange mobile GPUs in a similar manner. We weren’t briefed on the updates, most likely because there’s not much to say. Like AMD there are three "new" 600M parts. Here’s the overview of what NVIDIA is offering, with the previous generation equivalents listed for reference.

NVIDIA GeForce GT 635M, GT 630M, and 610M Specifications
  GeForce GT 635M GeForce GT 555M GeForce GT 630M GeForce GT 540M GeForce 610M GeForce 520MX
Core Name GF106/GF108 GF106/GF108 GF108 GF108 GF119 GF119
Stream Processors 144/96 144/96 96 96 48 48
Texture Units 24/16 24/16 16 16 8 8
ROPs 24/4 24/4 4 4 4 4
Core Clock 675/753MHz 675/753MHz 672MHz 672MHz 900MHz 900MHz
Memory Clock 1.8/3.6GHz DDR3/GDDR5 1.8/3.14GHz DDR3/GDDR5 Up to 900MHz (1.8GHz) DDR3 900MHz (1.8GHz) DDR3 900MHz (1.8GHz) DDR3 900MHz (1.8GHz) DDR3
Memory Bus Width 192/128-bit 192/128-bit 128-bit 128-bit 64-bit 64-bit
Memory Bandwidth 43.2/57.6GB/s 43.2/50.2GB/s 28.8GB/s 28.8GB/s 14.4GB/s 14.4GB/s

NVIDIA has the specifications up for their 600M parts, and it appears that they’ll be doing a straight rebadge without changing the clock speeds from the 500M equivalents—in fact, they’ll even keep the craziness that is the GT 555M. The only difference we could find is that GT 635M GDDR5 variants may have slightly more memory bandwidth (or more likely is that the spec page just doesn't adequately describe the bipolar nature of the product). What they will be changing is the apparent positioning of the products. The GT 630M and 610M drop 10 points from the model number, while the GT 635M drops 20 points; that appears to leave room for future GT 640M/650M parts, though nothing has been announced as yet. We also don’t have information on pricing, but there’s a possibility that with the drop in model number the prices will also be lower.

Like the AMD 7000M launch, GeForce 600M looks to be more about marketing and product positioning than anything. Mobile GPUs are about a generation behind their desktop counterparts, so with the renaming both AMD and NVIDIA are paving the way for new high-end GPUs to replace the current HD 6990M and GTX 580M. Thus, when we see the desktop HD 7970 and GTX 680 (or whatever they end up being named), we’ll should also see HD 7970M and GTX 680M. If recent history holds, those will end up being mobile variants of HD 7700 and GTX 660 (whatever those entail).

Introducing AMD’s Radeon Mobility 7400M, 7500M, and 7600M


View All Comments

  • erple2 - Friday, December 9, 2011 - link

    No, the 8800GT to 9800GT to GTS250 were the exact same chip, just with some die shrinkage. The 3870 was a substantial departure from what the 2900 series was - the die shrink alone wasn't enough to justify the substantial reduction in die size. Cutting back the 512bit memory bus cut it down a HUGE amount, too. There were also several minor tweaks to the architecture to boot. There was more redundant parts of the GPU in the core, too. Read up a bit more on it at http://www.anandtech.com/show/2679 ... Reply
  • chizow - Friday, December 9, 2011 - link

    Yes the R600 to RV670 shrink cut down the internal ringbus and external bandwidth in half which accounted for ~30M fewer transistors but *FUNCTIONALLY* the part was identical and just a rebadge/shrink of R600.

    Also, how can all the G92 variants be the "exact same chip" when you've already acknowledged a die shrink, not to mention the 8800/9800GT only had 112SP enabled where the 9800GTX/GTS 250 had the full 128SP enabled? You must have a different definition of "exact" i suppose. There's more differences between the products that I won't even bother getting into.

    Again, its ironic that certain people are so willing to overlook the differences in ATI's rebadges so conveniently forgets the differences with Nvidia's rebadges while condemning them, which is again, why people shouldn't bother nitpicking over such trivial matters.

    All that should matter to someone when buying is price, performance, and features and in all the cases of the various rabadges all 3 were just where they should've been relative to newer parts.
  • Sunagwa - Wednesday, December 7, 2011 - link

    Yea pretty much it right there.

    It's a sad day for gamers everywhere. On the bright side the real parts are right around the corner. =D
  • aguilpa1 - Wednesday, December 7, 2011 - link

    Wow, so pathetic it is beyond words... Reply
  • mariush - Thursday, December 8, 2011 - link

    They're forced by the laptop manufacturers - "if you guys don't come up with some new models, we'll go with the new models from nVidia".

    What would you want them to do?
  • evilspoons - Monday, December 12, 2011 - link

    I think this is why Apple doesn't list model numbers for this sort of thing for their laptops (both CPU and GPU, with the exception of the Intel HD 3000 for some reason - probably because it's so much better than the old Intel GPUs).

    For better or for worse, all they list is the RAM available (which I think is silly, since it's not as big a factor as the average person thinks it is) and the relative performance to their other models.
  • evilspoons - Monday, December 12, 2011 - link

    I stand corrected, they DO list the model number on the specs page... well, at least they don't bother with the ridiculous Intel CPU model number. It would be nice if they'd list turbo boost speeds though. Reply
  • lanestew - Wednesday, December 7, 2011 - link

    Saw the headline. Got excited. Then felt sad and disappointed. Reply
  • ganjha - Wednesday, December 7, 2011 - link

    Same here... Reply
  • piroroadkill - Wednesday, December 7, 2011 - link

    Wow. Yeah, same. I was about to get stuck in to lovely Southern Islands details, but what I got was a pile of re-heated mush.

    Poor show TSMC, poor show AMD.

Log in

Don't have an account? Sign up now