Subject: Graphics Cards | June 21, 2016 - 09:22 PM | Scott Michaud
Tagged: nvidia, fermi, kepler, maxwell, pascal, gf100, gf110, GK104, gk110, GM204, gm200, GP104
Techspot published an article that compared eight GPUs across six, high-end dies in NVIDIA's last four architectures: Fermi to Pascal. Average frame rates were listed across nine games, each measured at three resolutions:1366x768 (~720p HD), 1920x1080 (1080p FHD), and 2560x1600 (~1440p QHD).
The results are interesting. Comparing GP104 to GF100, mainstream Pascal is typically on the order of four times faster than big Fermi. Over that time, we've had three full generational leaps in fabrication technology, leading to over twice the number of transistors packed into a die that is almost half the size. It does, however, show that prices have remained relatively constant, except that the GTX 1080 is sort-of priced in the x80 Ti category despite the die size placing it in the non-Ti class. (They list the 1080 at $600, but you can't really find anything outside the $650-700 USD range).
It would be interesting to see this data set compared against AMD. It's informative for an NVIDIA-only article, though.
Subject: Graphics Cards | January 20, 2016 - 08:26 PM | Scott Michaud
Tagged: nvidia, linux, tesla, fermi, kepler, maxwell
It's nice to see long-term roundups every once in a while. They do not really provide useful information for someone looking to make a purchase, but they show how our industry is changing (or not). In this case, Phoronix tested twenty-seven NVIDIA GeForce cards across four architectures: Tesla, Fermi, Kepler, and Maxwell. In other words, from the GeForce 8 series all the way up to the GTX 980 Ti.
Image Credit: Phoronix
Nine years of advancements in ASIC design, with a doubling time-step of 18 months, should yield a 64-fold improvement. The number of transistors falls short, showing about a 12-fold improvement between the Titan X and the largest first-wave Tesla, although that means nothing for a fabless semiconductor designer. The main reason why I include this figure is to show the actual Moore's Law trend over this time span, but it also highlights the slowdown in process technology.
Performance per watt does depend on NVIDIA though, and the ratio between the GTX 980 Ti and the 8500 GT is about 72:1. While this is slightly better than the target 64:1 ratio, these parts are from very different locations in their respective product stacks. Swapping the 8500 GT for the following year's 9800 GTX, which leads to a comparison between top-of-the-line GPUs of their respective times, and you see a 6.2x improvement in performance per watt versus the GTX 980 Ti. On the other hand, that part was outstanding for its era.
I should note that each of these tests take place on Linux. It might not perfectly reflect the landscape on Windows, but again, it's interesting in its own right.
Maxwell and Kepler and...Fermi?
Covering the landscape of mobile GPUs can be a harrowing experience. Brands, specifications, performance, features and architectures can all vary from product to product, even inside the same family. Rebranding is rampant from both AMD and NVIDIA and, in general, we are met with one of the most confusing segments of the PC hardware market.
Today, with the release of the GeForce GTX 800M series from NVIDIA, we are getting all of the above in one form or another. We will also see performance improvements and the introduction of the new Maxwell architecture (in a few parts at least). Along with the GeForce GTX 800M parts, you will also find the GeForce 840M, 830M and 820M offerings at lower performance, wattage and price levels.
With some new hardware comes a collection of new software for mobile users, including the innovative Battery Boost that can increase unplugged gaming time by using frame rate limiting and other "magic" bits that NVIDIA isn't talking about yet. ShadowPlay and GameStream also find their way to mobile GeForce users as well.
Let's take a quick look at the new hardware specifications.
|GTX 880M||GTX 780M||GTX 870M||GTX 770M|
|GPU Code name||Kepler||Kepler||Kepler||Kepler|
|Rated Clock||954 MHz||823 MHz||941 MHz||811 MHz|
|Memory||Up to 4GB||Up to 4GB||Up to 3GB||Up to 3GB|
|Memory Clock||5000 MHz||5000 MHz||5000 MHz||4000 MHz|
Both the GTX 880M and the GTX 870M are based on Kepler, keeping the same basic feature set and hardware specifications of their brethren in the GTX 700M line. However, while the GTX 880M has the same CUDA core count as the 780M, the same cannot be said of the GTX 870M. Moving from the GTX 770M to the 870M sees a significant 40% increase in core count as well as a jump in clock speed from 811 MHz (plus Boost) to 941 MHz.
Subject: Graphics Cards | May 29, 2012 - 08:56 PM | Ryan Shrout
Tagged: nvidia, kepler, gtx 680, GTX 580, GK104, fermi
During the annual investors meeting with NVIDIA top-brass and the powers that hold the money on Wall Street, an interesting slide was presented in a somewhat veiled answer to the questions we (among many others) have had to the availability and yields of their latest Kepler GPUs.
At first glance, the graph would seem to validate claims that the stock and shipping rate of the new GeForce GTX 680 is simply unable to keep up with higher than expected and higher than normal demand. The line on the bottom represents the GTX 580 (both lines are mislabeled as GT rather than GTX) launch, the top the GTX 680 with the lower axis represented as weeks after launch. The vertical axis is labeled as "Units Sold Out Globally" but there are no numbers attached to it, making things incredibly vague.
When I asked for clarification all I was really given was that "it means sales of boards from AICs to distributors, system builders, e-tailers and retailers." This indicates that we are talking about boards either on Newegg.com, at Fry's or being sold through system builders like Maingear and Puget Systems.
NVIDIA's GTX 690 - one of the Kepler based cards MIA
The term "sold out" gave me a bit of pause - but when questioned "is it fair to translate 'units sold out globally' to 'units sold globally'?" I received an affirmative.
If we take NVIDIA's information as it is presented, then we see that after six weeks of product availability, the GTX 680 has sold and shipped at a rate 60% higher than that of the GeForce GTX 580 which launched in November of 2010. If that is true then we can agree with NVIDIA that demand is much higher for the GTX 680 than any other flagship GPU launch in recent memory and the continued stock and availability problems for Kepler are buyer created and strictly an NVIDIA yield issue.
NVIDIA has told me many times that they would obviously love to have more GTX 680s to sell to consumers as having them out of stock is only costing themselves money. How long it will take NVIDIA to balance out capacity with demand has yet to be seen though, so you can continue to check out our sort-of-weekly updates on GPU stock.
Subject: Graphics Cards | May 15, 2012 - 09:26 PM | Jeremy Hellstrom
Tagged: zotac synergy, zotac, nvidia, gt 630, gt 620, GT 610, GK104, geforce, fermi
Zotac has released ten different graphics cards today, three GT 630s, three GT 620s and four GT 610s if you count the PCI version.
Enjoy all the benefits of the new Kepler architecture without the price of the GTX 680 or 690. These cards are a mix of GF108, GF119 and GK107, essentially rebrands of previous GT series cards as opposed to new ones like NVIDIA would prefer you believed.
They range from this 4GB GT 630 Synergy Edition which will give you the ability to handle multiple monitors in a work environment.
To this 1GB PCIe 1x GT 610 model for low power, low profile applications where a 16x slot just won't fit.
HONG KONG – May 15, 2012 – ZOTAC International, a global innovator and channel manufacturer of graphics cards, mainboards and mini-PCs, today expands the successful ZOTAC GeForce 600 series with new value offerings. The ZOTAC GeForce GT 630, GT 620 and GT 610 series deliver a savory taste of Microsoft DirectX 11 technologies for an outstanding visual computing experience.
“ZOTAC is pleased to bring the GeForce 600 series to value shoppers seeking a superior visual experience discrete graphics brings to computing,” said Carsten Berger, marketing director, ZOTAC International. “By installing one of our ZOTAC GeForce GT 630, GT 620 or GT 610 series graphics cards, users can experience faster video and image processing and perfect high-definition video playback with a simple upgrade.”
The ZOTAC GeForce GT 630, GT 620 and GT 610 series are available in a variety configurations with 512MB, 1GB, 2GB and 4GB memory options in PCI Express 2.0 x16, PCI Express x1 or PCI interfaces, and active or passive cooling configurations to cater exclusively to all user computing needs.
It’s time to play with ZOTAC and the GeForce GT 630, GT 620 and GT 610 series.
- ZOTAC Expands successful GeForce 600 series
- ZOTAC GeForce GT 630 series
- 96 processor cores
- 1GB, 2GB and 4GB memory configurations
- 128-bit memory interface
- ZOTAC GeForce GT 620 series
- 96 processor cores
- 1GB & 2GB memory configurations
- 64-bit memory interface
- ZOTAC GeForce GT 610 series
- 48 processor cores
- 512MB, 1GB & 2GB memory configurations
- 64-bit memory interface
- NVIDIA 3D Vision capable
- NVIDIA Adaptive Vertical Sync
- DirectX 11 technology & Shader Model 5.0
- OpenGL 4.2 compatible
- Hardware-accelerated Full HD video playback
- Blu-ray 3D ready
- Loss-less audio bitstream capable
Subject: Graphics Cards | April 6, 2012 - 12:37 AM | Tim Verry
Tagged: rebranded gpu, nvidia, gtx 620, gtx 605, gpu, fermi
NVIDIA is continuing the infuriating practice (though they aren't alone in doing so) of re-branding older graphics cards into the following generation to present “new” cards (or to confuse consumers and tech journalists to no end, though I suspect that’s just a side benefit). Specifically, they are taking two lower tier desktop OEM cards and rebranding them as 600 series "Kepler" cards. The NVIDIA GT 520 and GT 510 will be renamed the NVIDIA GT 620 and GT 605 respectively. Even more confusing is that the “new” cards will have less hardware, and the only addition is the support for the OpenGL 4.2 standard (versus 4.1 on the 520 and 510). Other than that, they are two Fermi based cards in Kepler clothing.
The NVIDIA GT 620 replaces the GT 520 and features half of the graphics memory as the 500 series card, meaning that users will get 512 MB or 1 GB on the 620 instead of the 1 GB / 2 GB options of the GT 520. The card still features VGA, DVI, and HDMI video outputs. The remaining specifications can be seen in the chart below. Despite halving the memory, the new card has a very slightly higher TDP at 30 watts versus the rated 29 watts of the GT 520.
On the other hand, the NVIDIA GT 605 is the new version of the GT 510. The 600 series part also halves the amount of memory of the GT 510 counterpart with 512 MB and 1 GB versions compared to 1 GB and 2 GB versions of the GT 510. The GT 605 also has VGA, DVI, and HDMI ports. It is rated at a TDP of 25 watts like the GT 510.
|GT 510||GT 605||GT 520||GT 620|
|Graphics Clock||523 MHz||523 MHz||810 MHz||810 MHz|
|Processor Clock||1046 MHz||1046 MHz||1620 MHz||1620 MHz|
|Memory Clock (up to)||898 MHz||898 MHz||898 MHz||898 MHz|
|Memory (DDR3)||1 or 2 GB||512 or 1024 MB||1 or 2 GB||512 or 1024 MB|
Reference GT 500 and GT 600 Series Specifications (changes in bold).
According to Tom’s, the “new” cards are still Fermi based despite the new implied Kepler generation naming scheme. Granted, these are OEM cards but it still is a bit dishonest to rebrand them, especially in the case of the GT 620 where it is the same rank but with the Kepler generation digit at the beginning. There have been some comments around the Internet that the two new rebranded cards were brought into play to allow OEMs to sell PCs with new 600 series discrete graphics. At this level, it really doesn’t matter per se as they will still do HTPC and desktop graphics well enough and are not going to be purchased by customers directly, but it’s still annoying (heh). What do you guys think about the graphics card rebranding in general, whether it’s on the desktop or mobile market?
Subject: General Tech, Graphics Cards | March 22, 2012 - 01:51 AM | Jeremy Hellstrom
Tagged: pelly wants revenge, nvidia, leak, gtx 680, fermi, 28nm
Not to be outdone by NCIX, NewEgg also managed to jump the gun on the GTX 680 earlier today. The screengrab that was sent tp Bright Side of News shows their pricing of the soon to be released GTX 680 with models ranging from $500 to $535. The specs are there for all to see, a GPU running at 1.006GHz, Shader clock of 2.012GHz, effective memory of 6.008GHz and 1536 Stream Processors. Contrast that with the last GTX 580 that Josh reviewed which had a 782 MHz core, 1.564GHz shader, memory at 4.008GHz and 512 SPs and you can see it is a big step up!
If you visit NewEgg now you will be greeted with a different result, a page describing the GTX 680's various features and a Buy Now button which unfortunately doesn't work at this moment. That is a situtation which obviously cannot last as NewEgg would not have put it up. Of course the realization that you can pick up a pair of GTX 570's for the same price might just mean some recalculations will be in order once we see the performance of the actual card.
Subject: General Tech | March 21, 2012 - 03:26 AM | Jeremy Hellstrom
Tagged: pelly, nvidia, leak, gtx 680, fermi, 28nm
The gang over at Tweaktown managed to get pictures of a retail Gigabyte GTX 680, which is not only better than candid snaps from Las Vegas making it to the web, it also solidifies a few facts. For instance, as you can see below there are two 6-pin PCIe power connectors which pegs the maximum supplimental power that this card can draw at 150W. That is a big difference from the two 8-pin PCIe connectors that could deliver up to 275 to a GTX 580; NVIDIA has obviously made a huge step forward in power savings with the move to 28nm regardless of any design or manufacturing problems they may have had to overcome to deliver this card to retailers.
Tweaktown didn't stop there either GPU fans; it seems that the online\brick and mortar computer chain NCIX made a little mistake and let the GTX 680 appear on their wishlist app. Both an EVGA and an MSI model of the GTX 680 could be added to your wishlist ... for the price of $578.20 USD plus delivery. That same retailer currently sells HD 7970's for between $564.99 USD to $619.99. If only there had been some leaked benchmarks which might indicate which way AMD might have to adjust their pricing.
Lucas is so going to sue you!
Subject: General Tech | March 13, 2012 - 04:15 PM | Jeremy Hellstrom
Tagged: TSMC, nvidia, macbook, kelper, Ivy Bridge, fermi, apple
NVIDIA has been having a rough life lately with problems besetting them on all sides. Their IGP business has been disembowelled by AMD's Llano and even Intel is now offering usable graphics with the HD3000 on higher end Sandy Bridge chips. The console makers seem to have decided on AMD as the provider of choice for the next generation of products which locks NVIDIA out of that market for years to come, as console generations tend to last significantly longer than PC components. The delays at TSMC have enabled AMD to launch three families of next generation GPU without NVIDIA being able to respond, which not only hurts NVIDIA's bottom line but lets AMD set their own pricing until NVIDIA can finally release Kepler, at a price that will not be wholly of their choosing.
Now according to SemiAccurate they are losing a goodly portion of Apple's MacBook business as well. The supply issues which will be the result of the fabrication problems were likely a big factor in Apple's decision to trim back GPU orders but there is also the fact that the low to mid range GPU could well be going extinct. With the power of the forthcoming Intel HD4000 and AMD's Trinity line of APUs it will become hard for laptop and system makers to justify putting in a discrete GPU since they will have to choose relatively expensive parts to have the discrete GPU contribute to performance. That leaves NVIDIA only providing GPUs for high end MacBooks, a much less lucrative market than the mid range. Don't even mention the previous issue of overheating GPUs.
"That is exactly what SemiAccurate moles are telling us is going on. Nvidia can’t supply, so Apple threw them out on their proverbial magical experience. This doesn’t mean that Nvidia is completely out at Apple, the Intel GPUs are too awful to satisfy the higher end laptops, so there will need to be something in those. What that something is, we don’t definitively know yet, but the possibilities are vanishingly small."
Here is some more Tech News from around the web:
- Microsoft demos 1ms response touchscreen @ The Inquirer
- Asus: We are NOT killing off Transformer Prime @ The Register
- Diamond helps graphene carry more current @ Nanotechweb
- ASUS RT-N66U Wireless-N900 Gigabit Router @ Benchmark Reviews
- More CeBit Coverage @ XSReviews
Quarter Down but Year Up
Yesterday NVIDIA released their latest financial results for Q4 2012 and FY2012. There was some good and bad mixed in the results, but overall it was a very successful year for NVIDIA.
Q4 saw gross revenue top $953.2 million US with a net income of $116 million US. This is about $53 million less in gross revenue and $62 million down in net income as compared to last quarter. There are several reasons as to why this happened, but the majority of it appears to be due to the hard drive shortage affecting add-in sales. Simply put, the increase in hard drive prices caused most OEMs to take a good look at the price points of the entire system, and oftentimes would cut out the add-in graphics and just use integrated.
Tegra 3 promises a 50% increase in revenue for NVIDIA this coming year.
Two other reasons for the lower than expected quarter were start of the transition to 28 nm products based on Kepler. They are ramping up production on 28 nm and slowing down 40 nm. Yields on 28 nm are not where they expected them to be, and there is also a shortage of wafer starts for that line. This had a pretty minimal affect overall on Q4, but it will be one of the prime reasons why revenue looks like it will be down in Q1 2013.