Category:

GPU

NVIDIA RTX 30 Series – Everything You Need To Know

by Robin Sunny September 13, 2020

written by Robin Sunny

Finally NVIDIA unveiled the NVIDIA RTX 30 series cards in their keynote on the first day of September, after being teased about it for so long especially with the ‘Ultimate Countdown’. Promising us the biggest generational leap ever in the history of NVIDIAs gaming GPUs. The new Ampere architecture consumer cards are as follows:

NVIDIA Geforce RTX 3070
NVIDIA Geforce RTX 3080
NVIDIA Geforce RTX 3090

This new series is said to provide double the performance and power efficiency as the previous generation NVIDIA RTX 20 series. The series comes with next-generation of RT core, the fastest memory currently available i.e., GDDR6X and even more tensor cores for faster A.I. performance.

The new generation is also packed to the boot with software for the casual Gamers, Streamers and Creators, which we will go into later. To read about the earlier leaks about the NVIDIA RTX 30 series cards click here.

NVIDIA RTX 30 Series Specifications

NVIDIA had made some substantial claims about the performance of the latest GPU, but we were still overwhelmed after seeing the specifications of the current third-tier of the NVIDIA RTX 30 series GPUs i.e., the RTX 3070 when its compared to the flagship NVIDIA RTX 2080Ti.

Graphic Card	CUDA cores	RT Cores	Tensor Cores	Memory	Memory interface	Boost clock	Graphics card power
RTX 3090	10496	256	1024	24GB GDDR6X	384 bit	1.70GHz	350W
RTX 3080	8704	144	576	10GB GDDR6X	320 bit	1.71GHz	320W
RTX 3070	5888	112	448	8GB GDDR6	256 bit	1.73GHz	220W
RTX 2080 Ti	4352	88	544	11GB GDDR6	352 bit	1.54GHz	250W
RTX 2080 Super	3072	64	384	8GB GDDR6	256 bit	1.82GHz	250W
RTX 2070 Super	2560	64	320	8GB GDDR6	256 bit	1.77GHz	215W

Comparison Of The RTX 30 Series And RTX 20 Series

These are still just the official details provided by NVIDIA (although the benchmarks have been leaked online), the official benchmarks have yet to be released. Although these aren’t the first Ampere GPU to show up in the market.

NVIDIA RTX 3070

The NVIDIA RTX 3070 was introduced alongside the RTX 3080 and RTX 3090 on the September 1st keynote event, but in spite of that the card will not be released along with the other cards. The NVIDIA RTX 3070 will be released on October 2020 with the MSRP of $499.

Out of the cards that were introduced in the keynote, the RTX 3070 is the cheapest and on par (even slightly better on paper) compared to the earlier high-end NVIDIA RTX 2080Ti. Considering its price which is less than half of the 2080Ti.

But the thing that might hold the RTX 3070 back from being the best 4K gaming GPU is the 8GB GDDR6 especially with the 256-bit Bus. However, this might be the best gaming GPU for those who want to game in 4K while on a budget.

NVIDIA RTX 3080

With the NVIDIA RTX 3080, the story might be a bit different compared to the RTX 3070. With more CUDA Cores, Tensor Cores, higher Memory Capacity and speed, the power consumption spikes along with that.

The RTX 3080 goes from around 220W of the RTX 3070 to a kookie 320W. NVIDIA doubled the CUDA Cores in the RTX 3080 to an insane 8704 from the 4352 CUDA Core in the RTX 2080 Ti.

The card is scheduled to be released on 17 September at $699 MSRP, which is actually way better than the estimated prices that were being thrown around prior to the keynote. Although still a bit pricey, but considering the huge leap in performance, it isn’t as bad as we thought.

But what we feel odd is how NVIDIA bought down the memory from 11GB to 10GB, considering RTX 3080 is the successor of the previous RTX 2080 (Yes, we realise the memory is faster).

NVIDIA RTX 3090

If you thought that the NVIDIA RTX 3080 was bonkers, check out the specs for the RTX 3090 in the above table. Along with that 350W TDP comes the epic 10496 CUDA Cores powering this beast leaving the NVIDIA RTX 2080 Ti far behind in performance. Even the Titan RTX falls short in-front of this behemoth of a card, boasting 256 RT Cores and 1024 Tensor Cores.

During the Keynote even the NVIDIA Co-Founder Jensen Huang called this card the ‘Big Ferocious GPU’ or ‘BFGPU’ in short.

NVIDIA announced that the RTX 3090 would come out after RTX 3080 in September 2020. Pricing-wise, we were expecting Team Green to follow its own trend of increasing the price vastly every new generation, but fortunately and to our surprise the MSRP wasn’t the estimated around $2000 but $1500.

Yes, its still one of the most expensive graphic cards on the market, but given its huge leap in performance, it was not as expensive as everyone predicted. That does not make it cheap, but to buy this card at launch you better have been saving up some dough.

That Weird Cooler Design

In case there are still people who haven’t seen the new cooler design that NVIDIA has put into its RTX 30 series cards, these were even leaked a few months ago, here it is again:

The card has a fan on both sides (one on top and one on the bottom) of that massive heatsink which spans the entire length of the card. Even the card is designed in a weird way to accommodate this heatsink.

Due to this design there is some chatter in the community if the hot air blown out of the GPU will increase the thermals of a CPU AIO mounted on top of the case chassis due to hot being sucked up by the AIO unnecessarily increasing the CPU temperatures, but we will see how this fares as time goes go and we have more information on this card. This heatsink will only be available for the RTX 3080 and RTX 3090 and as for the RTX 3070, it will feature the good o’l tried and tested dual fan setup we all are used to.

NVIDIA Reflex

In the Keynote NVIDIA reflex was also introduced, focused as an E-Sports technology to dramatically reduce latency. It was explained that Reflex optimizes the rendering pipeline across CPU and GPU to reduce latency by up to 50%, it’s to be released along with the newer NVIDIA drivers. Keeping up with the theme of E-Sports, four new 360Hz G-Sync Gaming monitors were also announced from Asus, Alienware, Acer and MSI.

NVIDIA Broadcast

During the Keynote Jensen Huang spoke of some of the feature that were coming out along with RTX 30 series cards one among them was the NVIDIA Broadcast. It was shown as an AI assisted feature that could help streamers on a budget.

NVIDIA Broadcast had features like Noise Cancellation, Video effects and more. Here is the video below if you want check it out for yourself.

NVIDIA Omniverse Machinima

As stated by Jensen Huang “Omniverse is a universal design tool asset exchange, with a viewer based on photo-realistic path tracing. The engine is designed to be physically accurate simulating light, physics, material, and artificial intelligence” or in other words you can bring in assets to make a photo-realistic cinematic experience, with the help of AI to make face animations with your voice and extremely accurate physics simulations to help a user create a short animated videos or to help make movies. The Beta to this application starts this October.

NVCache

This is a piece of technology that was rumored to be in development, the way that a CPU loads data onto a GPU is by copying it on the system memory and then decompressing it onto the GPU,this is how it used to work but now with much faster PCIe Gen 4 SSDs PCs can transfer file much quicker, now the bottleneck is the part where a CPU copy’s data into the system memory. But with this the storage directly loads the data onto the GPU memory thereby reducing load times. The GPU now dynamically uses the RAM, VRAM and SSD to complete multiple tasks much faster than what was previously possible.

DLSS 2.0

Deep Learning Super Sampling (DLSS) was announced in the NVIDIA Keynote as a way to play your favorite RTX games at a higher resolution with minimal performance penalty. DLSS was unveiled with four games that are going to receive DLSS through patches. These games are:

Control
Deliver Us The Moon
MechWarrior 5: Mercenaries
Wolfenstein: Youngblood

Benchmarks

Now there are leaked benchmarks of the RTX 3080 from Videocardz that have been verified by them and just as we expected the new 30 series card out performs the previous generations enthusiast cards. Here are the benchmarks:

September 13, 2020 1 comment

Computer GPU Nvidia

NVIDIA To Cease Production Of It’s RTX 20 Series Lineup

by Robin Sunny July 16, 2020

written by Robin Sunny

NVIDIA has reportedly shut down the production of it RTX 20 series cards for the new Ampere Architecture cards set to come out soon, most likely to push customers to buy the new upcoming lineup.

Due to this, we might start seeing the prices of the ‘Turing’ GPU’s to rise even higher due pandemic and the sudden mining craze that is being revived.

According to Ithome, NVIDIA might start to suspend the production of it’s high-end RTX 20 series cards due to TSMC’s insufficient production quantity. The subsequent high end cards like the RTX 2080 and the RTX 2080 Ti might be out of stock soon.

The RTX 20 series lineup that are having their production halted are as follows :

GeForce RTX 2070
GeForce RTX 2070 SUPER
GeForce RTX 2080 Ti
GeForce RTX 2080 SUPER

All cards utilize different Turing GPU, we get the TU102 on the GeForce RTX 2080 Ti, the TU104 on GeForce RTX 2080 & RTX 2070 SUPER and the TU106 GPU on the GeForce RTX 2070. The report also claims that not only would the production be halted on these four graphics cards but they will also be de-listed by major vendors including NVIDIA themselves in the coming months.
Those who are looking to get a Turing graphics card right now will be disappointed by the fact that a price hike on the GeForce RTX 20 series is being talked around by several retailers across the globe. That and with the resurgence of the mining craze is going to end up inflating prices on all existing high-end cards. The price hike is suggested to occur this month & ‘My Drivers’ report that even the mainstream GeForce GTX 1660 SUPER & the GeForce RTX 2060 will be affected. It is stated that NVIDIA has internally notified its AIB partners to adjust prices with the demand accordingly.
Wccftech

We are not sure how reliable this news is just yet, and the mining craze would probably die down just as soon as it is being bought back to life. But this price hike of the RTX 20 series cards might just be a silver lining for the next generation Ampere lineup. During the last generation, large quantities of Pascal GPUs were prepped due to the mining craze and several OEMs and retailers had to price hike the initial Turing cards to clear away inventory of older cards. But now, if the stock of Turing cards are empty we might start seeing the upcoming card at MSRP (Manufacturer Suggested Retail Price) that NVIDIA suggested rather than the inflated prices retailers used earlier to clear the old GPU’s.

But as of now most PC builders do not have an option other than to wait out these price hikes and hope that the prices of upcoming NVIDIA lineup don’t cost an arm and a leg.

July 16, 2020 0 comment

Computer GPU Nvidia

NVIDIA’s Ampere Lineup with the RTX 3070/3080/3080 Ti. Significantly More Faster than the Current Top Tier GPU’s

by Robin Sunny July 2, 2020

written by Robin Sunny

With the rumours floating around Nvidia’s next gen GPU with it’s brand spanking new architecture Ampere, along with news of the alleged fact that NVIDIA has been discussing Ampere and the RTX 3000 series cards with it’s AIB partners, there are quite a few tid-bits that might be amazing to see if its true.

GeForce RTX 3000 Series

The 3000 series will have “HUGE” performance improvements in Ray Tracing due to its newer Ampere architecture which gives it faster ray tracing than the older Turing architecture, along with being more power efficient. Rasterization will also receive a big upgrade allowing for more realistic looking worlds in games when paired with Ray Tracing.

The GeForce RTX 3000 series card will also have significantly more VRAM than other consumer cards so we might just see:

GA102-400 – 24GB
GA102-300 – 12GB
GA102-200 – 10GB

We might also see an increase in clock speeds compared to the Turing cards in around the range of 100/150-200MHz, while having better power efficiency thanks to Ampere and its new 7nm process.NVIDIA will be using the 7nm EUV (Extreme ultraviolet lithography) process node for Ampere, something that will deliver NVIDIA some amazing power savings.

But the Intersting part is that NVIDIA has had really great power efficiency since Pascal architecture, and went right into Turing with improved power efficiency and that’s on larger 14nm and 12nm nodes. 7nm gives NVIDIA much more room to play in, joined with the Ampere GPU architecture. Rumor has it the new Ampere GPUs can run at under 1.0v which is absolutely huge and could mean for the most power efficient GeForce cards yet.

Although it might be wise to take these rumours/speculations with a grain of salt until we get closer to a time period where we would actually expect a new NVIDIA GPU Arch to release. Many rumours might be inaccurate or contain half-truths.

NVIDIA had been slowly but steadily preparing the launch of their Ampere generation GPUs starting from March 2020 starting with its high-end cards like the A100. While the 3000 series cards might take a while to release (around August or September-ish) on the consumer PC’s thanks to the pandemic.

There have been various graphics card names thrown around for the cards mentioned above, its a speculation that it might fall under these names:

GeForce RTX 3090 or RTX Titan
GeForce RTX 3080 Ti or 3080 Super
GeForce RTX 3080

Benchmarks

The prominent “leaker” Rogame (@_rogame) found a private listing on 3DMark, which he claims to have identified as being uploaded by an NVIDIA employee and this is how the card stacks up:

The 6000MHz memory on the ‘Unknown’ NVIDIA is probably due to 3D Mark (the software that was used for this benchmark) not having no data on the type of memory. With the cards having a 1935MHz clock speed.

Specifications aside, this mystery card blows the stock RTX 2080Ti out of the water performing 31% better than the current top gaming card (although 2080 Ti is not specifically for gaming, it falls between the line of professional/content creation and gaming chores) the exception being EVGA’s overclocked Kingpin card.

This is very impressive even if we consider this as the top-end card, things are looking good for Ampere on the consumer side. We also have to consider that the performance will most probably increase from this point on (assuming this leak is legit) as NVIDIA’s driver team polishes the rough edges.

Possible Design Leak

The leaked image appears to feature the RTX 3080 Founders Edition and its interesting fan design. From the photos, it looks like the card will feature two fans and while that’s nothing out of the ordinary, it should be noted that the fans are positioned on the opposite sides of the heat sink.

We don’t know how this will influence airflow just yet but we’re certain that Nvidia has already addressed this and know what they are doing. It does, however, raise the issue of cost as this design definitely costs extra.The accepted speculation sets the price point of this design at around $150 and that definitely adds fuel to the fire of some rumours guessing that the overall price of this next gen card would be even steeper than the previous generation.

What makes these leaks credible is the fact that Nvidia itself announced that it will be making an internal investigation to the nature of those leaks, thereby confirming their validity. It’s said that these images got leaked before even product and sales managers got to see them.

July 2, 2020 1 comment

Computer GPU Nvidia

Why the New NVIDIA Ampere A100 Isn’t Made For You

by Robin Sunny May 18, 2020

written by Robin Sunny

Nvidia announced its new GPU, the A100 a few days ago. Although calling it a GPU might be a bit of a stretch as this ‘GPU’ is mainly focused towards A.I. Computing and Scientific Computing, still it is based on the Ampere architecture so it might give us an hint about GeForce gaming card that may come out next.

Now graphics cards by their nature are designed to do a limited set of operations really quick and in parallel like the Million plus polygons your typical games uses in each frame, because of this people also realised that you can process a huge amount of data much more faster than a CPU and thus began the use of GPUs in servers and super-computers. And that is exactly what this A100 card is intended for.

Nvidia plans to sell this in their DGX-1 systems as sort of a pod that links 8 of these systems together and can be slotted into a server and sells for $199,999. But Nvidia says it would be worth it considering the performance the A100 can deliver.

The A100 is based on the Ampere architecture and it’s also manufactured on the 7nm process, now moving to to a lower nanometer process from its earlier 12nm process should give them better energy efficiency and enable them to cram in more transistors on each chip and that’s exactly what they did, Nvidia claims that the A100 packs 54 Billion transistors into a 826 sq. millimetre dye, for context Nvidia’s current gaming card the RTX 2080ti has 18.6 Billion transistors in a 754 sq. millimetre dye. This can potentially mean that the A100 packs as much as 3 times the performance as this top of the line gaming card.

Comparing the A100 to a RTX 2080ti might be the wrong call here, Nvidia has really set the A100 to be the successor of the V100 its other Data center GPU released in 2018, being a very impressive GPU in its own right the V100 retailed for around $9,000. The V100, although plenty fast, Nvidia claims that the newer A100 will be as much as 20 times faster than its predecessor. Of course all this processing wouldn’t mean much if it couldn’t move data fast enough, but with 14GB of HBM2 (High Bandwidth Memory) capable of moving 1.5TB per second i.e., 40 times faster than a High End CPU server, it should be fine.

In addition to the 8 A100 GPUs the DGX system also sports a 15TB Gen4 NVME flash storage, a Terabyte of RAM and 2 64-core AMD Rome CPUs. So what does this card tell us about the next graphics cards? Well the Ampere architecture has been positioned as the successor of the Volta architecture, but Volta never spawned any consumer cards and the current RTX series graphic cards are based on the Turing architecture, but Volta did introduce Tensor cores that are specialised for A.I. math that did find its way into consumer cards. And it is possible that the A.I. improvements seen in the Ampere card would also make its way into the gaming cards, helping with A.I. based Upscaling like DLSS or in De-Noising that is used in Ray Tracing.

So will it be possible to get your hands on new GPU as fast as the A100? Probably not, but with the technologies used in this card we might start seeing future cards getting significantly faster with more transistors and better A.I. for that sweet Ray tracing.

May 18, 2020 3 comments

Computer CPU GPU

2nm Chips in the Works from TSMC, Expected to Release by 2025

by Robin Sunny May 2, 2020

written by Robin Sunny

A few days ago TSMC, one of the leading semiconductor foundry, announced the R&D of a new 2nm chip. The efficiency and performance of a Chipset i.e., your CPU or GPU, depends on how many transistors can be accommodated on a Chipset and the only way to accommodate those transistors is to make the Chipsets bigger or to make the transistors smaller. To put it in perspective how small the new transistor will be, a single strand of human hair is 75,000nm in diameter. So 2nm is just around 10 atoms wide. The current best generation of processors use the 7nm process with 5nm coming soon.

Being in very early stages of development there is not much information besides its existence. This year we might start seeing companies like Intel and AMD start adopting the 7nm and we hope to see the first 5nm chips, so expect to see big improvements in the coming future.

May 2, 2020 1 comment

Computer GPU

GTX 1660 Super – The Best Budget 1080p Gaming GPU

by Robin Sunny April 28, 2020

written by Robin Sunny

Alright, to be completely honest this is just a glorified GTX 1660 with GDDR6 memory, but this makes big difference in its performance which I would go into later.

The GeForce GTX 1660 Super, launched in October 2019 and based on a 12nm process with the Turing architecture. Although, unlike the 1660 Ti, which uses the same GPU and has all 1536 shaders unlocked, Nvidia disabled some of the shading units to reach the targeted shader count of 1408. Now for the memory upgrade, the GTX 1660 went through a massive change, from GDDR5 with 192GB/s to GDDR6 with 336GB/s, before this card was launched, the only other card with this fast memory was the RTX 2060.

Now the average Joe might ask “What’s the difference between GDDR5 and GDDR6?” and “Is it worth shelling out the extra cash?”, the answer to latter is YES, the increase in the memory bandwidth means the card can draw frames faster and draw higher quality images faster, having a low memory bandwidth can mean the GPU is suddenly bottlenecked, as it’s waiting for the RAM to send it data, suddenly limiting the frames or FPS it can spit out.

Source: TechRadar
Source: TechRadar
Source: TechRadar

The GeForce GTX 1660 Super retailes for around ₹21,000 (as of Q1 2020), while the GTX 1660 Ti is around ₹26,500 (as of Q1 2020), the 2%-3% increase in performance doesn’t exactly validate its extra ₹5,000.

If you are in the market for a 1080p gaming graphics card, I would recommend buying the GTX 1660 Super. The original GTX 1660 was already a very good entry level GPU, Nvidia made it better in every single way. The Nvidia Super series cards have been a bit of a hit and miss (I’m looking at you RTX 2080 Super), but I can’t recommend GTX 1660 Super enough.

At the end of the day, 1080p is the most popular gaming resolution in the PC section, and graphic cards of this price range is where people will end up spending their money.

April 28, 2020 0 comment