NVIDIA 400M: DX11 Top to Bottom Solutions Now Available
by Jarred Walton on September 3, 2010 12:02 AM ESTPerformance and Mainstream 400M
After the high-end parts, the drop in performance can become precipitous. This has been particularly bad for AMD GPUs, where the drop from Mobility 5800 series down to the 5700 and 5600 parts often means less than half the performance. NVIDIA has had a few more upper-midrange parts floating around, though, and that looks to continue.
| NVIDIA Performance and Mainstream 400M Specifications | |||||
| GeForce GT 445M | GeForce GT 435M | GeForce GT 425M | GeForce GT 420M | GeForce GT 415M | |
| Codename | GF106 | GF108 | GF108 | GF108 | GF108 |
| CUDA Cores | 144 | 96 | 96 | 96 | 48 |
| Graphics Clock (MHz) | 590 | 650 | 560 | 500 | 500 |
| Processor Clock (MHz) | 1180 | 1300 | 1120 | 1000 | 1000 |
| Memory Clock (MHZ) | 800/1250 | 800 | 800 | 800 | 800 |
| Standard Memory Configuration | DDR3/GDDR5 | DDR3 | DDR3 | DDR3 | DDR3 |
| Memory Interface Width | 128/192-bit | 128-bit | 128-bit | 128-bit | 128-bit |
| Memory Bandwidth (GB/sec) | 25.6/60.0 | 25.6 | 25.6 | 25.6 | 25.6 |
| SLI Ready | No | No | No | No | No |
First, you'll notice that none of these "Performance and Mainstream" parts supports SLI. That's hardly surprising, as SLI with lower-end mobile GPUs has never been our recommended approach. First get to the high-end for performance reasons, and then worry about SLI. Other than that limitation, all of these parts have the same features as the faster parts on the previous page.
The new GT 445M is the first part to come with split specifications. Given the option for 128-bit and 192-bit bus widths, it appears the 445M will use the full GF106 memory controller for the higher bandwidth version and cut off one of the 64-bit interfaces for the low bandwidth model. Many of our gaming results have looked bandwidth limited, so we'd definitely recommend going for the GDDR5 192-bit model if possible, but that will be up to the notebook manufacturers. 445M looks to compete in a similar space as 460M with the higher bandwidth model, but it cuts computational power quite a bit at roughly two-thirds of the 460M. The difficulty here is that 445M can be either substantially faster than some of the older parts, or if you get the 128-bit DDR3 model you're suddenly cut down to less than half the bandwidth. Heavy use of shaders, tessellation, etc. might make the lack of bandwidth less painful, but without hardware and future games it's difficult to say how things will play out.
The 435M is a more straightforward replacement of GT 335M. (Did someone ask for a remake of M11x with a DX11 GPU? Hopefully they can do something about the LCD this time around….) 335M has 72 cores at 1080MHz, with 34.1GB/s of bandwidth. Unless something changes, 435M will actually have less bandwidth but substantially more computational power—60% more to be exact (plus architectural changes, obviously). This is a pattern that holds throughout the 400M lineup, so NVIDIA appears to be betting heavily that shader performance rather than bandwidth will become important.
Along with the 435M come several more GPUs; the 425M and 420M have the same bandwidth and core counts, but lower core/shader clocks. This is similar to the current 325M/330M, which have 48 cores but the same amount of bandwidth as the 335M. Even the lowest 420M has around 25% more compute power than 335M, but they all have less bandwidth. It would have been nice to see a move to GDDR5 on more of the Performance and Mainstream parts, as that would have improved overall performance substantially.
Finally, wrapping up the low end we have the GT 415M. Here we can actually see something to celebrate, since the previous generation parts largely consisted of 16 core models with a 64-bit bus (i.e. the G 310M). On the compute side, we're looking at nearly twice the power of the G 310M. Bandwidth also gets a kick in the pants, going from 12.8GB/s to 25.6GB/s. In short, our entry-level mobile GPUs just doubled their performance. Note also that if NVIDIA wanted to cuts things down even further, they'd need to make yet another chip (i.e. GF110), since 48 cores is a single SM. Most likely, for anything below GT 415M they'll just continue to sell their older 300M parts.
Facebook
Twitter
RSS
39 Comments
View All Comments
Roland00 - Friday, September 3, 2010 - link
The reason I ask is for this is the most popular "mainstream" gaming card in current designs of laptops right now. Yes the GTX 480m may be the fastest but if nobody uses it besides select Alienwares and Clevos what does it matter?JarredWalton - Friday, September 3, 2010 - link
The GT 335M is roughly at the same performance level as HD 5650, but obviously without DX11 support. Based on that, I would venture that even 420M might match the 5650; certainly 425M and up will be faster (unless you're bandwidth limited, which is entirely possible).marraco - Friday, September 3, 2010 - link
The CPU integrated video of Intel will rule out all the "entry level" discrete chips.Nvidia and AMD will be forced to offer more juicy entry level cards.
I guess that the next generation of entry level will have at least geforce 8600 power.
JarredWalton - Friday, September 3, 2010 - link
Our initial look suggest that the 12 EU version of the Sandy Bridge IGP will be at roughly the level of the G 310M. In other words, even the GT 415M looks to be around twice as fast, plus you get CUDA, OpenCL, DirectCompute, OpenGL 4.0, DX11, PhysX, etc. I'd say Sandy Bridge will make a direct replacement for G 310M (i.e. "G 410M") pointless, so perhaps that's why there's no castrated 24 core 400M chip with a 64-bit interface.tviceman - Friday, September 3, 2010 - link
Looks like a solid lineup from everything below the gtx480m. Hopefully like others have suggested they updated the 480m to be a little more powerful at the same TDP or offer the same performance at a lower TDP.I'm personally still waiting on any word or significant rumors regarding a a full 384 core GF104 desktop part.
blah238 - Friday, September 3, 2010 - link
Most interested in the 12-14" size range with the fastest possible GPU and decent screen options. The Sony Vaio Z is the only machine that fits the bill currently but it's way out of my price range.Here's hoping something from this refresh gets stuck into a chassis that gives Sony some competition in this space.
patrickjchase - Saturday, September 4, 2010 - link
Jarred makes reference to the fact that the GF1xx series strike a different balance between compute (shader) and memory bandwidths. I think that part of NVIDIA's motivation is indeed an expectation that future games will skew towards requiring more shader performance, but I think there's another factor: Fermi has an L1 data cache in each SM.I do OpenCL and CUDA programming professionally, and I think that it's important not to underestimate the impact of cache. Many algorithms in graphics and elsewhere have access patterns that are best described as "localized but unpredictable". This means that the algorithm's data accesses tend to "cluster" spatially and/or temporally, but it's very difficult to predict *where* they'll cluster and it's therefore impractical to explicitly pre-load data into local memory.When running such algorithms Fermi needs less DRAM bandwidth for any given performance level than any other GPU on the market (and again, I say this as somebody who develops for and benchmarks these things day in and day out).
This is actually a bit of a repeat of how general-purpose CPUs and their associated memory systems progressed, beginning with the IBM 360/85 all the way back in 1969...
JackNSally - Monday, September 6, 2010 - link
"OpenGL 40." in the supported features. I believe you mean "OpenGL 4.0" unless nVidia is jumping far, FAR into the future.JarredWalton - Monday, September 6, 2010 - link
The Alienware M15x has already done 1080p in a 15.4" chassis I think, and there are several other 1080p ~15" laptops around. ASUS G51JX-X1, ThinkPad W510, and MSI's GX660R-060US are all 15.6" 1080p. But then, they're also non-3D and cost $1300 minimum.