NVIDIA Gives RTX A6000 "Ada" Professional Graphics a Quiet Launch, Starting $7377

[btarunr]

NVIDIA is ready to launch its RTX A6000 series "Ada" professional-visualization graphics cards. These cards are targeted at the same market demographic as the NVIDIA Quadro series of the old—serious 3D content creation. The RTX A6000 leads the pack, and is based on the 4 nm "AD102" silicon (the same one powering the GeForce RTX 4090). The A6000 is better endowed than the RTX 4090 at the silicon-level, although operating at lower GPU clock-speeds, for its tighter 300 W power-limit (compared to 450 W of the RTX 4090).

The A6000 "Ada" is endowed with 18,176 CUDA cores across 142 SM, compared to the 16,384 CUDA cores across 128 SM of the RTX 4090. It also gets a higher number of Tensor cores, at 568. The defining differentiator between the A6000 and RTX 4090 has to be memory, with the pro-vis card getting 48 GB of ECC GDDR6 memory across the chip's 384-bit memory bus, clocked at 20 Gbps (960 GB/s memory bandwidth); compared to the 24 GB of 21 Gbps GDDR6X (1008 GB/s) of the RTX 4090. Also, the card enables all three NVDEC and NVENC video hardware-accelerators physically present on the AD102, for six independent accelerated transcoding streams.



The card features a 16-pin ATX 12VHPWR connector, and probably includes a 3x 8-pin to 16-pin adapter, as the card's typical board power is just 300 W. The reference board features a 2-slot design with a high-density channel-type heatsink with vapor-chamber plate; and a lateral-flow blower. There is no NVLink support. Leadtek, PNY, and ELSA are expected to be the board partners for this card. Both PNY and Leadtek put out pricing, with the PNY card being cheaper at USD $7,377.



View at TechPowerUp Main Site | Source

 

[Arco]

 

[Nater]

Titan please!

I don't want to run dual systems. Or even dual video cards.

 

[MentalAcetylide]

Now this looks like a nice card for rendering and only takes up two slots. I have a Lenovo RTX A6000 from the previous generation and it is suppose to be better than the other brands, such as PNY. I wonder which is the better quality out of the bunch this time around? I know PNY is pretty much a "generic" brand, but not sure about Leadtek or ELSA. I'll have to find out which brand government research institutes are going to be using for workstations since Lenovo doesn't appear to be making these workstation cards this time around. It will be good to know in the future if & when I need to upgrade.

However, it does suck that there's no NVLink support. While I'm nowhere near using 48Gb VRAM for rendering at present, it would be nice to have the NVLink option on future generations in case the need for more pooled VRAM is needed. I guess there just isn't much of a demand for NVLink so they decided to cut it.

 

[bonehead123]

$7.3k+...... hahahahahaha

Yea these cards may be "latest & greatest" etc, but the old Quadro cards were not nearly as much when they launched, even today, the last 48GB version was only ~$5.3k.....

nGreedia seems to have upgraded their pharma to the next level......

 

[MentalAcetylide]

$7.3k+...... hahahahahaha

Yea these cards may be "latest & greatest" etc, but the old Quadro cards were not nearly as much when they launched, even today, the last 48GB version was only ~$5.3k.....

nGreedia seems to have upgraded their pharma to the next level......

Not during the crypto craze! They were selling for around $6000, and that was for the crap PNY brand. The only models that I saw for that price before they went into orbit was PNY, which is garbage. I think one of the reasons(in part) that Nvidia is jacking up its prices is so it has more $$$ to dump into buying up manufacturing resources... and maybe get Huang a few more closets filled with leather jackets.

 

[Zareek]

Now this looks like a nice card for rendering and only takes up two slots. I have a Lenovo RTX A6000 from the previous generation and it is suppose to be better than the other brands, such as PNY. I wonder which is the better quality out of the bunch this time around? I know PNY is pretty much a "generic" brand, but not sure about Leadtek or ELSA. I'll have to find out which brand government research institutes are going to be using for workstations since Lenovo doesn't appear to be making these workstation cards this time around. It will be good to know in the future if & when I need to upgrade.

However, it does suck that there's no NVLink support. While I'm nowhere near using 48Gb VRAM for rendering at present, it would be nice to have the NVLink option on future generations in case the need for more pooled VRAM is needed. I guess there just isn't much of a demand for NVLink so they decided to cut it.

ELSA was originally a computer hardware brand from Germany. They had a very solid reputation. They went bankrupt twenty years ago, and after that it is a bit confusing.

 

[InVasMani]

Not during the crypto craze! They were selling for around $6000, and that was for the crap PNY brand. The only models that I saw for that price before they went into orbit was PNY, which is garbage. I think one of the reasons(in part) that Nvidia is jacking up its prices is so it has more $$$ to dump into buying up manufacturing resources... and maybe get Huang a few more closets filled with leather jackets.

You're not thinking big enough the more you buy the more Huang saves for the missile silo to fill with leather jackets.

 

[Gaslongjumping71]

Might as well buy 4 4090's at that price point.

 

[Selaya]

wake me up in 2024 when they're releasing the RTX AD2000, won't ya

 

[WhoDecidedThat]

Full AD102 clocked at 2500 MHz and it's rated for 300 watts? How?!

 

[Icon Charlie]

ELSA was originally a computer hardware brand from Germany. They had a very solid reputation. They went bankrupt twenty years ago, and after that it is a bit confusing.

Damn! That does bring back memories.

 

[mechtech]

Price and model not over 9000!!!

pass

 

[sepheronx]

You're not thinking big enough the more you buy the more Huang saves for the missile silo to fill with leather jackets.

He plans to build a ICBJ, first of its kind, in order to one up Best Koreas leader?

 

[CGLBESE]

Now this looks like a nice card for rendering and only takes up two slots. I have a Lenovo RTX A6000 from the previous generation and it is suppose to be better than the other brands, such as PNY. I wonder which is the better quality out of the bunch this time around? I know PNY is pretty much a "generic" brand, but not sure about Leadtek or ELSA. I'll have to find out which brand government research institutes are going to be using for workstations since Lenovo doesn't appear to be making these workstation cards this time around. It will be good to know in the future if & when I need to upgrade.

However, it does suck that there's no NVLink support. While I'm nowhere near using 48Gb VRAM for rendering at present, it would be nice to have the NVLink option on future generations in case the need for more pooled VRAM is needed. I guess there just isn't much of a demand for NVLink so they decided to cut it.

Damn... last time I got a Leadtek card in my hands, it was their A250TDH... Hardly heard from them since, which is a bit sad. Their products were quite well engineered, with pretty good cooling, nice overclocking room and a well thought software suite.

Are they still present on European markets?

 

[dir_d]

Does this come with SR-IOV or GRID for VDI desktops at this price point?

 

[DemonicRyzen666]

Full AD102 clocked at 2500 MHz and it's rated for 300 watts? How?!

HBM vs GDDR6x maybe ?

 

[Jism]

$7.3k+...... hahahahahaha

Yea these cards may be "latest & greatest" etc, but the old Quadro cards were not nearly as much when they launched, even today, the last 48GB version was only ~$5.3k.....

nGreedia seems to have upgraded their pharma to the next level......

Price for a wafer rised; inflation, the whole thing. However as a pro card thats where the money is at for both AMD and Ndivia.

 

[Dr. Dro]

Full AD102 clocked at 2500 MHz and it's rated for 300 watts? How?!

It throttles under heavy load, and the GDDR6 memory uses significantly less energy as well.

The card should sustain decent clocks throughout, though, even the 4090 is known for using a lot less than its TDP under regular load (unlike the 3090, which constantly hits its power limit).

 

[MentalAcetylide]

Might as well buy 4 4090's at that price point.

Very true + it would be a LOT faster, but for the iray rendering I do, if a scene exceeds 24Gb VRAM, it won't matter how many 4090s I have since it wouldn't fit on any of the cards and would just end up getting dropped to the cpu. So in a sense, you're paying extra for a much larger chunk of VRAM.
Plus I don't think there are any motherboards big enough with appropriately spaced slots to support 4x 4090s and the huge custom water cooling setup that would probably be needed to prevent too much heat from building up in the case. So we're stuck with having to buy more 2-slot workstation cards if we desire more CUDA cores for more speed and need 3-4 of them to fit on the same board without any space issues. This could be one of the possible reasons behind Nvidia making the gaming cards ridiculously huge & power hungry in comparison to their workstation cards.

 

[Karti]

HBM vs GDDR6x maybe ?

these are on GDDR6 non-x, not HBM

 

[DemonicRyzen666]

these are on GDDR6 non-x, not HBM

oh thanks, but then it's just GDDR6x Vs GDDR6. Doesn't really explain the power difference.

 

[Σario]

Very true + it would be a LOT faster, but for the iray rendering I do, if a scene exceeds 24Gb VRAM, it won't matter how many 4090s I have since it wouldn't fit on any of the cards and would just end up getting dropped to the cpu. So in a sense, you're paying extra for a much larger chunk of VRAM.
Plus I don't think there are any motherboards big enough with appropriately spaced slots to support 4x 4090s and the huge custom water cooling setup that would probably be needed to prevent too much heat from building up in the case. So we're stuck with having to buy more 2-slot workstation cards if we desire more CUDA cores for more speed and need 3-4 of them to fit on the same board without any space issues. This could be one of the possible reasons behind Nvidia making the gaming cards ridiculously huge & power hungry in comparison to their workstation cards.

I run comino waterblocks for my a6000s and they will for sure make them for this card. Single slot with quick disconnects. Board is a Gigabyte MC62-G40.

You can easily fit multiple 4090s with single slot blocks and a manifold configuration. It's too bad there's not more professional products on the market instead of kiddy RGB junk.

 

[MentalAcetylide]

I run comino waterblocks for my a6000s and they will for sure make them for this card. Single slot with quick disconnects. Board is a Gigabyte MC62-G40.

You can easily fit multiple 4090s with single slot blocks and a manifold configuration. It's too bad there's not more professional products on the market instead of kiddy RGB junk.

Yeah, but then the maintenance really starts to become a problem for me. That's just way too much for someone like me to handle with regards to custom water cooling setups.

 

[TheoneandonlyMrK]

Clearly were at max pharma levels, surely someone's nose is melting in Nvidia's pricing department.
Nice card though.

Though that naming scheme wow, very poor imho.