Quote:The GTX 340 is perhaps one of the strangest and rarest graphics cards you can find online today. As covered by the Budget-Builds Official YouTube Channel, the GTX 340 is, in fact, an unofficial GeForce 300 series card that Nvidia never actually produced. Instead, enterprising modders created the cards from standard GT units by adding souped-up dual-slot coolers, modifying the BIOS, applying a hefty overclock, and then changing the identifier string to make the card appear as a "GTX" model in the system.
...
So what about the GTX 340? Apparently, someone years and years ago got into the business of upgrading GT 340s, which were then re-sold to system builders. The mods included a dual-slot cooler, a modified BIOS with the GTX nomenclature, and a hefty core overclock. In fact, the card shows up as a GTX 340 in GPU-z.

This strategy does make sense — the GT 340 is already very power efficient, consuming just 51W, so getting a good overclock out of the GT 340 wasn't hard at all with a cooler that's twice the size of the original.

Unfortunately, we don't know why this person (or people) decided to give the GT 340 the GTX 340 badge. Still, it was probably to help differentiate these heavily modified cards from the traditional GT 340.
...
Of course, with how old this card is, performance is very underwhelming; the card can barely manage 30FPS in CSGO at 480P (yeah, it's that bad), and the performance delta between the GT 340 and the "GTX 340" is barely 2%, at most.

That's the story of the GTX 340; it's perhaps one of the most boring GPUs ever to be released, but it's cool that someone made a brand new SKU out of "thin air," and it seemed to have worked for them. At least from a business perspective. But good luck trying to buy one — they are super scarce on eBay and can only be bought overseas.

Quote:As spotted by HotHardware, Anthony, a hardware enthusiast from the Mod Labs forums, has recreated 3dfx Interactive's renowned Voodoo 5 6000. The company never released the Voodoo 5 6000 to the public, but the skillful enthusiast managed to revive the fallen graphics card through some exquisite reverse engineering work.
...
Anthony's creation isn't exactly a faithful copy of the original Voodoo 5 6000, but some may argue that it looks even better. Instead of the old-school green PCB, he created his own black PCB, which adds a bit of a modern look to the graphics card. He equipped it with four VSA-100 dies, which cost $18.95 apiece, with their corresponding heatsinks and cooling fans.

The Voodoo 5 6000 was certified for a TDP of 60W, which is more than the AGP slot can provide. Therefore, it drew its power through an external 250W power brick. Anthony artfully added a standard 4-pin Molex power connector to his rendition of the Voodoo 5 6000 to run the graphics card with a common computer power supply. Putting aside these small changes, Anthony assures that his Voodoo 5 6000 performs the same as the original because he replicated the same BIOS, drivers, and even the same bugs.

Aside from the high production cost, 3dfx didn't launch the Voodoo 5 6000 due to several bugs with the AGP x4 slot on some motherboards. As a result, the graphics card was forced to run at x2. Some of the reported issues included color distortion and artifacts in games, translucent stripes on the screen, and other miscellaneous bugs.

Quote:The Transaction Processing Performance Council (TPC), an industry consortium that develops performance benchmarks for servers and data centers and includes names like AMD, Intel, and IBM, has accused Nvidia of violating the TPC’s Fair Use Policy. To compare the performance of its GPUs against other machines, Nvidia presented results from workloads derived from official TPC benchmark workloads and compared them to official TPC results, a violation of the group's fair use policy.

At its GTC Fall 2020 virtual event last year, Nvidia presented a paper called State of RAPIDS: Bridging the GPU Data Science Ecosystem. In the paper, Nvidia demonstrated the advantages of its RAPIDS suite of GPU-accelerated software libraries for data science workflows over competing hardware and software platforms.
...
The GPU giant claimed that its platform "outperformed by nearly 20x the record for running the standard big data analytics benchmark, known as TPCx-BB." However, Nvidia didn't run the workloads designed by TPC and used similar workloads instead, which violates TPC's policies and makes such comparisons invalid and generally unfair. However, Nvidia did mark the version of the TPCx-BB benchmark it used as 'not official.'

"The TPC actively encourages publicizing of TPC results by all organizations, including the press, market researchers, financial analysts and non-profit organizations," said Mike Brey, chairman of the TPC Steering Committee. "However, to ensure that users and readers of TPC results are given a fair and complete representation of TPC data, the TPC requests that all users follow the Fair Use rules, outlined in TPC policies, when publishing or publicizing results."

The TPC said it was working with Nvidia to correct the issue.

TPC is led by companies that produce CPUs, servers, and business software, including AMD, IBM, and Intel. It also includes cloud data center operators like Alibaba, Microsoft, and Oracle. Nvidia is not a member of the TPC.

Quote:Nvidia's Ampere architecture has just fairly recently arrived to consumers (at least for those who've managed to find it in stock), but that hasn't stopped leakers from figuring out what's next on Nvidia's agenda, at least for its future gaming architectures. 3dCenter.org Tweeted on new rumored specifications for Nvidia's future "Lovelace" AD102 chip, which might include 144 SMs and 18432 CUDA cores with around 66 TFlops of computing performance.

Beware, this is entirely based on rumors, and with how little we know about Nvidia's next GPU architectures, there's a huge chance most of this info is wrong. But, at least this post can give us some clues as to where Nvidia is headed.

One part of this rumor that does make sense is the codename "Lovelace"; Jensen a few years back at CES of 2018 was wearing a T-shirt with several names of popular mathematicians all through history. And with Nvidia's love for codenaming architectures with names of mathematicians, many believe Jensen was leaking the names of future architectures on his T-Shirt. One of those names was Lovelace.
...
If you were wondering about Nvidia's other rumored architecture, Hopper, it seems right now that both Hopper and Lovelace are in the works. What makes Hopper interesting is its multi-die design, which would be a first for Nvidia. What we could see is Lovelace being Nvidia's gaming architecture, as one die decreases latency, then we could see Hopper as being a data center exclusive.

Quote:The rumor mill has begun grinding with details about NVIDIA's next-gen graphics processors based on the "Lovelace" architecture, with Kopite7kimi (a reliable source with NVIDIA leaks) predicting a 71% increase in shader units for the "AD102" GPU that succeeds the "GA102," with 12 GPCs holding 6 TPCs (12 SMs), each. 3DCenter.org extrapolates on this to predict a CUDA core count of 18.432 spread across 144 streaming multiprocessors, which at a theoretical 1.80 GHz core clock could put out an FP32 compute throughput of around 66 TFLOP/s.

The timing of this leak is interesting, as it's only 3 months into the market cycle of "Ampere." NVIDIA appears unsettled with AMD RDNA2 being competitive with "Ampere" at the enthusiast segment, and is probably bringing in its successor, "Lovelace" (after Ada Lovelace), out sooner than expected. Its previous generation "Turing" architecture saw market presence for close to two years. "Lovelace" could leverage the 5 nm silicon fabrication process and its significantly higher transistor density, to step up performance.

Quote:ASRock has introduced its first graphics card in an M.2-2280 form-factor. The M_2 VGA module is designed to add legacy display connectivity to any system that has an M.2 slot with a PCIe interface.

The ASRock M_2 VGA module uses Silicon Motion's SM750 display controller with 16MB of embedded memory and a PCIe x1 interface. The card has a 15-pin header to connect a D-Sub output that supports resolutions of up to 1920×1440. Since this is a single-chip board, its power consumption is rated at 1.49 Watts, which is almost negligible by today's standards in the PC world. Meanwhile, the unit still has an additional power connector.

The Silicon Motion SM750 is a rather simplistic display controller that only supports 2D graphics and a basic video engine. The chip features two display engines and has two 300 MHz RAMDACs, one TMDS transmitter, and one LVDS transmitter.
...
ASRock Rack traditionally does not disclose prices of its products. Given that the M2_VGA module is to a large degree a niche device, it will be available only from resellers specializing in this kind of products.

Quote:If you're a programmer you'd have /facepalm'd by now, let me explain. In a multi-threaded program, Events are often used to synchronize concurrently running threads. Events are a core feature of the Windows operating system, once created, they can be set to "signaled", which will notify every other piece of code that is watching the status of this event—instantly and even across process boundaries. In this case the Radeon Settings program will wait for an event called "DVRReadyEvent" to get created, before it continues with initialization. This event gets created by a separate, independent, driver component, that's supposed to get loaded on startup, too, but apparently never does. The Task Scheduler entries in the screenshot above do show "StartDVR". The naming suggests it's related to the ReLive recording feature that lets you capture and stream gameplay. I guess that part of the driver does indeed check if Radeon hardware is present, and will not start otherwise. Since Windows has no WaitForEventToGetCreated() function, the usual approach is to try to open the event until it can be opened, at which point you know that it does exist.

You're probably asking now, "what if the event never gets created?" Exactly, your program will be hung, forever, caught in an infinite loop. The correct way to implement this code is to either set a time limit for how long the loop should run, or count the number of runs and give up after 100, 1000, 1 million, you pick a number—but it's important to set a reasonable limit.
...
Waiting on synchronization signals is very basic programming skills, most midterm students would be able to implement it correctly. That's why I'm so surprised to see such low quality code in a graphics driver component that get installed on hundreds of millions of computers. Modern software development techniques avoid these mistakes by code reviews—one or multiple colleagues read your source code and point out potential issues. There's also "unit testing", which requires developers to write testing code that's separate from the main code. These unit tests can then be executed automatically to measure "code coverage"—how many percent of the program code are verified to be correct through the use of unit tests. Let's just hope AMD fixes this bug, it should be trivial.

If you are affected by this issue, just uninstall the AMD driver from Windows Settings - Apps and Features. If that doesn't work, use DDU. It's not a big deal anyway, what's most important is that you are aware, in case your system feels sluggish after a graphics hardware change.

Quote:In a surprising turn of events, Nvidia today announced that it's completely killed its current model of SLI, which lets your system run more than one Nvidia graphics cards simultaneously for greater performance.

Nvidia has transferred all SLI implementation responsibilities to the game developer and game engine and won't release new SLI driver profiles for the RTX 20-series and older GPUs starting January 1.

"For GeForce RTX 3090 and future SLI-capable GPUs, SLI will only be supported when implemented natively within the game," Nvidia said.
...
What does this mean for SLI? Nvidia hasn't done SLI any favors over the past several years. The GTX 900-series was the final architecture by Nvidia to fully support SLI in its entirety, specifically 3-way and 4-way SLI configurations.
...
Now, with Ampere, Nvidia has limited SLI to the RTX 3090, which is poised to be one of if not the best graphics card on the market upon release. But there's good reason for this. Since its inception, SLI has been riddled with bugs, crashes and inconsistent frame times, resulting in choppy gameplay. Getting the tech working correctly requires a lot of optimization from the game developer and from Nvidia. Additionally, there isn't a large market of PC gamers running more than one graphics card in their PC.

With Nvidia dropping SLI support for the RTX 3080 and below, Nvidia can free up resources, allowing its driver teams to work on other things. You might get performance improvements due to low-level API support. Nvidia has stated that SLI's only purpose right now is to make the fastest gaming rigs in the world, and this is a good way to keep that going.

Quote:Buried in the bowels of the driver change-log is a big disclosure, that the RTX 30-series "Ampere" GPUs don't support SLI (as in implicit multi-GPU), only explicit multi-GPU through modern APIs such as DirectX 12 and Vulkan (if supported by the game). NVIDIA has already restricted the NVLink interface needed for SLI only to its topmost RTX 3090 SKU, so the lack of implicit multi-GPU effectively marks the end of SLI for future GPUs. That said, the technology itself is very much present on all GPUs that support it, provided games do.

1. Doom Eternal massively favors AMD cards from 7 years ago vs. Nvidia cards from 7 years ago
   
2. Hardware Unboxed botched the testing
   
3. AMD bribed them
   

Quote:In its testing, Hardware Unboxed found that in pairs of GPUs from AMD and Nvidia, where they used to battle neck-in-neck at their release, AMD's GPUs today come out with significantly higher performance than Nvidia's counterparts in testing Doom Eternal.

Take for example the GTX 780 and its main competitor from when it was out seven years ago, the almighty AMD Radeon R9 290. Back then, the two battled closely with one or the other winning at different games, but in Doom Eternal we see something completely different. Whereas the GTX 780 is able to put down an average of 45 FPS, the R9 290, pushes out 116 FPS on average -- that's more than twice as fast as Nvidia's card!

Quote:AMD has released a statement regarding a theft of its graphics IP and what may have been a subsequent attempt to blackmail the company. The statement reads:

Quote:At AMD, data security and the protection of our intellectual property are a priority. In December 2019, we were contacted by someone who claimed to have test files related to a subset of our current and future graphics products, some of which were recently posted online, but have since been taken down.

While we are aware the perpetrator has additional files that have not been made public, we believe the stolen graphics IP is not core to the competitiveness or security of our graphics products. We are not aware of the perpetrator possessing any other AMD IP.

We are working closely with law enforcement officials and other experts as a part of an ongoing criminal investigation.

That’s all the detail we have right now, but these kinds of efforts rarely end well for the parties involved. AMD, Intel, Nvidia, and other semiconductor companies have often cooperated to prosecute individuals who have tried this kind of stunt before. While it might seem ideal for one company to attempt to get a leg up on another by directly copying its product design, this kind of trick is fraught with legal peril for the company that tries it.
...
We’ve reached out to various sources within AMD and several have quietly confirmed that the company is framing the situation accurately. The IP in question wasn’t core to either the competitiveness or the security of future AMD products.