Volta Discussion Thread

[SteelCrysis]

Let's start with this: https://www.techpowerup.com/224413/nvidi...o-may-2017

[RolloTheGreat]

Let's start with this: https://www.techpowerup.com/224413/nvidi...o-may-2017

LOL

The guys over at ATi are probably yelling, "FuuuuuuuuuuuuuuuuuuuuuuuuuuuckkkkkkkK!!!!! We are so fucking doomed!!!"

(Yes I know it is AMD now, but they will always be ATi to me)

[SteelCrysis]

http://wccftech.com/nvidia-volta-12nm-finfet/

Whispers floating around the tech-sphere allege that NVIDIA is in fact launching its next generation Volta GPUs this year on 12nm FinFET. This hot rumor, which Fox Business has picked up, claims that NVIDIA’s next generation family of GPUs will succeed Pascal this year in the GPGPU supercomputing, machine learning and artificial intelligence spaces. Furthermore, Volta is said to be built using a custom, cutting edge, 12nm FinFET manufacturing process at TSMC.

[SteelCrysis]

http://techreport.com/news/31582/report-...nm-process

We're still gawping at the massive performance of the GeForce GTX 1080 Ti, but here's a nugget of news about Nvidia's next big thing. According to a report from Taiwan's Commercial Times, TSMC will be fabricating Nvidia's upcoming Volta GPUs on its new "12nm" process. Currently, TSMC's most efficient process is called "16nm FFC", and given that it debuted back in April 2015, it's about time for an update.

[SteelCrysis]

https://www.techpowerup.com/233187/nvidi...tesla-v100

This chip is a beast of a processor: it packs 21 billion transistors (up from 15,3 billion found on the P100); it's built on TSMC's 12 nm FF process (evolving from Pascal's 16 nm FF); and measures a staggering 815 mm² (from the P100's 610 mm².) This is such a considerable leap in die-area that we can only speculate on how yields will be for this monstrous chip, especially considering the novelty of the 12 nm process that it's going to leverage. But now, the most interesting details from a gaming perspective are the 5,120 CUDA cores powering the V100 out of a total possible 5,376 in the whole chip design, which NVIDIA will likely leave for their Titan Xv. These are divided in 84 Volta Streaming Multiprocessor Units with each carrying 64 CUDA cores (84 x 64 = 5,376, from which NVIDIA is cutting 4 Volta Streaming Multiprocessor Units for yields, most likely, which accounts for the announced 5,120.) Even in this cut-down configuration, we're looking at a staggering 42% higher pure CUDA core-count than the P100's. The new V100 will offer up to 15 FP 32 TFLOPS, and will still leverage a 16 GB HBM2 implementation delivering up to 900 GB/s bandwidth (up from the P100's 721 GB/s). No details on clock speed or TDP as of yet, but we already have enough details to enable a lengthy discussion... Wouldn't you agree?

[SteelCrysis]

https://www.techpowerup.com/240482/nvidi...ash-binary

NVIDIA could be giving final touches to its Quadro GV100 "Volta" professional graphics card, after the surprise late-2018 launch of the NVIDIA TITAN V. The card was found listed in the binary view of the latest version of NVFlash (v5.427.0), the most popular NVIDIA graphics card BIOS extraction and flashing utility. Since its feature-set upgrade to the TITAN Xp through newer drivers, NVIDIA has given the TITAN family of graphics cards a quasi-professional differentiation from its GeForce GTX family.

[SteelCrysis]

https://www.techpowerup.com/241500/nvidi...next-month

NVIDIA is reportedly working on a TITAN V-esque surprise for March 2018. According to Reuters, which summarized the company's Q4-2017 results and outlook, the company is working on a new consumer-graphics GPU for launch next month, codenamed "Turing." This could be the codename of an ASIC or an SKU and not the architecture (which could be "Volta").

[SteelCrysis]

Never mind, Turing is actually a cryptocurrency mining GPU: https://www.techpowerup.com/241552/nvidi...-pc-gaming

[RolloTheGreat]

https://www.techpowerup.com/241500/nvidi...next-month

NVIDIA is reportedly working on a TITAN V-esque surprise for March 2018. According to Reuters, which summarized the company's Q4-2017 results and outlook, the company is working on a new consumer-graphics GPU for launch next month, codenamed "Turing." This could be the codename of an ASIC or an SKU and not the architecture (which could be "Volta").

Year or two of "I'm waiting to see AMDs response, because I want to play upcoming game X/mine cryptocoin Y/buy monitor Z" at ATVF incoming.

Meanwhile, I ordered this G Sync monitor to use with my 1080Ti:

https://www.amazon.com/AOC-AG352UCG-3440...B06X9CBRTP

3440 X 1440 is pretty easy for the 1080Ti to handle, so I can ride out the Volta mining gouger prices using the 1080Ti.

[SickBeast]

https://www.techpowerup.com/241500/nvidi...next-month

NVIDIA is reportedly working on a TITAN V-esque surprise for March 2018. According to Reuters, which summarized the company's Q4-2017 results and outlook, the company is working on a new consumer-graphics GPU for launch next month, codenamed "Turing." This could be the codename of an ASIC or an SKU and not the architecture (which could be "Volta").

Year or two of "I'm waiting to see AMDs response, because I want to play upcoming game X/mine cryptocoin Y/buy monitor Z" at ATVF incoming.

Meanwhile, I ordered this G Sync monitor to use with my 1080Ti:

https://www.amazon.com/AOC-AG352UCG-3440...B06X9CBRTP

3440 X 1440 is pretty easy for the 1080Ti to handle, so I can ride out the Volta mining gouger prices using the 1080Ti.  

From what I'm hearing Navi is half baked and is basically Vega 2.0. AMD is completely screwed at this point. Now they're talking about the next architecture beyond Navi but I don't even know if they will make it that far.

IMO Vega was AMD's Parhelia. I miss Matrox. It was a proud Canadian company. Let's not forget that AMD's graphics division truly is also. I have finally put aside my Canadian pride and I support nVidia these days.

[RolloTheGreat]

https://www.techpowerup.com/241500/nvidi...next-month

NVIDIA is reportedly working on a TITAN V-esque surprise for March 2018. According to Reuters, which summarized the company's Q4-2017 results and outlook, the company is working on a new consumer-graphics GPU for launch next month, codenamed "Turing." This could be the codename of an ASIC or an SKU and not the architecture (which could be "Volta").

Year or two of "I'm waiting to see AMDs response, because I want to play upcoming game X/mine cryptocoin Y/buy monitor Z" at ATVF incoming.

Meanwhile, I ordered this G Sync monitor to use with my 1080Ti:

https://www.amazon.com/AOC-AG352UCG-3440...B06X9CBRTP

3440 X 1440 is pretty easy for the 1080Ti to handle, so I can ride out the Volta mining gouger prices using the 1080Ti.  

From what I'm hearing Navi is half baked and is basically Vega 2.0.  AMD is completely screwed at this point.  Now they're talking about the next architecture beyond Navi but I don't even know if they will make it that far.

IMO Vega was AMD's Parhelia.  I miss Matrox.  It was a proud Canadian company.  Let's not forget that AMD's graphics division truly is also.  I have finally put aside my Canadian pride and I support nVidia these days.

I stepped up 1080 FTW to !080Ti FE when they launched, so I paid MSRP for 1080Ti. The ATVF goofballs would certainly say, "Wah! It self OCs a 100 or 200MHZ less than the three fan models!" but a. I got it pretty quick after launch b. I'm not an OCer and those 1-200MHz don't make much difference in reviews I've seen c. it's fast enough to run ULTRA setting on 3440X1440 on GSync 100Hz monitor from some Joker benches I saw.

AMD just doesn't have the jing to compete with NV any longer. They're "kind of" competing with Risen, and I'm glad, but by the time Navi launches there will be a 2080Ti that stomps it.

[SickBeast]

https://www.techpowerup.com/241500/nvidi...next-month

NVIDIA is reportedly working on a TITAN V-esque surprise for March 2018. According to Reuters, which summarized the company's Q4-2017 results and outlook, the company is working on a new consumer-graphics GPU for launch next month, codenamed "Turing." This could be the codename of an ASIC or an SKU and not the architecture (which could be "Volta").

Year or two of "I'm waiting to see AMDs response, because I want to play upcoming game X/mine cryptocoin Y/buy monitor Z" at ATVF incoming.

Meanwhile, I ordered this G Sync monitor to use with my 1080Ti:

https://www.amazon.com/AOC-AG352UCG-3440...B06X9CBRTP

3440 X 1440 is pretty easy for the 1080Ti to handle, so I can ride out the Volta mining gouger prices using the 1080Ti.  

From what I'm hearing Navi is half baked and is basically Vega 2.0.  AMD is completely screwed at this point.  Now they're talking about the next architecture beyond Navi but I don't even know if they will make it that far.

IMO Vega was AMD's Parhelia.  I miss Matrox.  It was a proud Canadian company.  Let's not forget that AMD's graphics division truly is also.  I have finally put aside my Canadian pride and I support nVidia these days.

I stepped up 1080 FTW to !080Ti FE when they launched, so I paid MSRP for 1080Ti. The ATVF goofballs would certainly say, "Wah! It self OCs a 100 or 200MHZ less than the three fan models!" but a. I got it pretty quick after launch b. I'm not an OCer and those 1-200MHz don't make much difference in reviews I've seen c. it's fast enough to run ULTRA setting on 3440X1440 on GSync 100Hz monitor from some Joker benches I saw.

AMD just doesn't have the jing to compete with NV any longer. They're "kind of" competing with Risen, and I'm glad, but by the time Navi launches there will be a 2080Ti that stomps it.

Well, IIRC my highly overclocked water cooled GTX 1080 Ti is around 8% faster than a stock Titan XP card. Overclocking does make a difference on these cards. But I'm sure your FE card is fine. If you were able to step up to it, good for you.

I think I gained around 20% over stock via overclocking with my card. But I really got a golden sample. It boosts to 2000mhz+ and I can get +1000mhz on the memory. I maxed out the slider in Afterburner. I've never done that before.

[RolloTheGreat]

https://www.toptengamer.com/gtx-1080-ti-...benchmark/

That guy got those benches with a FE OCd 200MHz, so even if you take 10% off (which is probably high for that OC) everything there is more than playable on a 100Hz GSync monitor.

(and for that matter if you add 10% to those benchmarks it makes no difference at all on a 100Hz GSync monitor)

So, won't be buying the watercooling kit, or OCing.

[SteelCrysis]

Quadro GV100, revamped Tesla V100 announced: http://www.tomshardware.com/news/nvidia-...36748.html

[SteelCrysis]

https://www.extremetech.com/extreme/2665...imulations

Nvidia has long held the pole position in GPGPU computing, particularly in scientific and HPC applications. The company’s long-term investment into CUDA and high performance computing have won it a number of spots in the supercomputing TOP500 and fueled the growth of its Tesla product line, including GPUs like the $3,000 Titan V, a Volta-based graphics card that straddles the line between a consumer and a scientific product. But all may not be well with the Titan V — there are reports that the chip can produce different results from run to run.

That’s the word from The Register, which writes:
...
It seems unlikely, however, that scientists and researchers would mistake a known issue (non-deterministic output in parallel FP calculations) for a significant hardware issue. The Reg’s source indicated the Titan V could give incorrect results about 10 percent of the time, but did not provide details on which applications were affected, whether the frequency of the problem varied from application to application, or if it could be impacted by changing various GPU settings.

Right now, what we have are more questions than answers. The problem, if it exists, might be addressable via driver or a code change. It might also reflect a problem with the GPU’s memory subsystem, as The Reg speculates. Some HPC applications have updated their own websites to indicate they are aware of the potential issue and haven’t seen it yet. It’s also possible that the issue is limited to a handful of cards and not indicative of a general problem.

As for Nvidia, the company has told the Reg it is aware of the issue and has invited anyone affected to contact Nvidia itself. The Titan V isn’t really positioned as a gaming GPU, but games do not appear to be impacted or affected at this time.

[SteelCrysis]

https://www.techpowerup.com/243979/next-...nd-of-2018
I'm guessing that this will be Volta. Be sure to read the rest, it's a short article.

An official Gigabyte UK Notebook representative, who goes by the name of Atom80, over at the OverclockersUK forums has confirmed that NVIDIA's next-generation mobile GPUs will launch towards the end of this year.

[SteelCrysis]

GTX 1180 sighted in TPU's GPU database: https://www.techradar.com/news/nvidia-ge...a-titan-xp

[SteelCrysis]

https://techreport.com/news/33835/nvidia...eural-nets

Nvidia CEO Jensen Huang loves to take the world by surprise with new Titan cards in unexpected places, and yesterday was no different. At the Computer Vision and Pattern Recognition conference running in Salt Lake City, Utah, Huang unveiled the Titan V CEO Edition. This beefed-up Volta card boasts 32 GB of HBM2 memory, a substantial boost from the Titan V's 12 GB of stacked RAM.
...
Nvidia gave away 20 of these cards to researchers at the conference and didn't announce general availability, so getting one will apparently require being chosen by Huang himself for the foreseeable future. Indeed, the company says that the 20 recipients at CVPR "won't be the last to be so honored." Nvidia also describes this card as a limited edition, so between the extra cache and cachet, any generally available version is likely to be even pricier than the $3000 the first Titan V commands—somewhere between that figure and the $9000 sticker of the Quadro GV100. We'll keep our eyes peeled for a product page and price tag.

[SteelCrysis]

https://www.techpowerup.com/245501/nvidi...tx-1180-ti

My take on this whole board, mostly due to the overkill power supply (up to 525 W) and the amount of test points and jumpers is that this board is used to test and qualify performance and power consumption in an unconstrained way, so that engineers and marketing can later decide on acceptable power and performance targets for release. The NVLink connector and functionality can also be tested at this stage, and the final PCB for mass production will be designed based on the outcome of these tests. On the bottom left of the PCB we find a mini-DP connector, which should be perfectly sufficient for this kind of testing, but not for a retail board.
...
Update:
I tried to estimate die size from the photo. We do know from the datasheet that the GDDR6 memory chips are 14 mm x 12 mm. Based on that information I rescaled, warped and straightened the image, so that each GDDR6 memory chip is 140 x 120 pixels. With all memory chips around the GPU now being at the correct size, we can use the GPU's silkscreen print to estimate the actual size of the chip package, which I measured at 48.5 x 48.5 mm. Assuming that the inner silk screen with the solder balls represents the surface of the GPU die, we get a length of 26 mm for each side of the die, which brings die size to 676 mm². This makes it a relatively large die considering NVIDIA's existing lineup: GV100 (815 mm², Titan V), GP100 (610 mm², Quadro GP100), GP102 (471 mm², GTX 1080 Ti), GP104 (314 mm², GTX 1080), GP106 (200 mm², GTX 1060). So my initial assessment that this could be the GP102 successor seems accurate, especially since GV100 die size is quite a bit bigger than GP100 die size, by roughly 33%. Our calculated GV102 die size is 40% bigger than GP102, which falls in that range.

[SteelCrysis]

Lenovo accidentally reveals that the next series of Geforce graphics cards is the 11-series: https://www.techpowerup.com/245580/lenov...-11-series

[SteelCrysis]

https://www.techpowerup.com/245606/due-t...gen-launch

Almost as an afterthought, DigiTimes also adds that NVIDIA has decided to postpone launch of their next-gen products (both 12 nm and then, forcibly, 7 nm) until supply returns to safe levels.