NVIDIA's Tiny RTX 4000 Ada Lovelace Graphics Cards is now Available

[AleksandarK]

NVIDIA has begun selling its compact RTX 4000 Ada Lovelace graphics card, offering GeForce RTX 3070-like performance at a mere 70 W power consumption, allowing it to fit in almost all desktop PCs. The low-profile, dual-slot board is priced higher than the RTX 4080 as it targets professional users, but it can still be used in a regular gaming computer. PNY's RTX 4000 Ada generation graphics card is the first to reach consumer shelves, currently available for $1,444 at ShopBLT, a retailer known for obtaining hardware before its competitors. The card comes with four Mini-DisplayPort connectors, so an additional mDP-DP or mDP-HDMI adapter must be factored into the cost.

The NVIDIA RTX 4000 SFF Ada generation board features an AD104 GPU with 6,144 CUDA cores, 20 GB of GDDR6 ECC memory, and a 160-bit interface. With a fixed boost frequency floating around 1560 MHz to reduce overall board power consumption, the GPU is rated for just 70 Watts of power. To emphasize the efficiency, this card requires no external PCIe power connector, as all the juice is fed through the PCIe slot. The GA104 graphics processor in this configuration delivers a peak FP32 performance of 19.2 TFLOPS, comparable to the GeForce RTX 3070. The 20 GB of memory makes the card more valuable for professionals and AI researchers needing compact solutions. Although the card's performance is overshadowed by the recently launched GeForce RTX 4070, the RTX 4000 SFF Ada's professional drivers, support for professional software ISVs, and additional features make it a strong contender in the semi-professional market. Availability and pricing are expected to improve in the coming weeks as the card becomes more widely accessible.



More images, along with specification table, follow.




View at TechPowerUp Main Site | Source

 

[P4-630]

@Tomgang might be a nice upgrade for your RTX A2000 6 GB...

 

[Bones]

Tiny?
I've got cards that only takes up 1 slot, not two like these do but at the same time I guess they are rather small for what's being sold these days.

 

[Tomgang]

@Tomgang might be a nice upgrade for your RTX A2000 6 GB...

Interesting card, but priced higher than a rtx 4080, while only perform like a rtx 3070. Not worth it.

I am all ready covered with rtx a2000 and a rtx 4090. So I'll pass on this one.

However a card with 8-10 gb vram and a lot cheaper. I could be interested. For now it's prised to high.

 

[Vayra86]

Tiny ADA - Two slot card.

 

[gffermari]

This is a weird card.
More CUDAs than a 4070, nearly twice the vram and performance like a 3070....
What the hell...
....and all that so they keep it below 70W....

 

[Tomgang]

This is a weird card.
More CUDAs than a 4070, nearly twice the vram and performance like a 3070....
What the hell...
....and all that so they keep it below 70W....

70 watt limit comes with a price to pay. Performance and core clock. with a gpu clock boost of just around 1560 mhz is very low for what ada lovelace is capable of, thinking on that up to 3 GHz is possible. However i am sure many would love a card like this, if the price was a lot lower. We need something better that these low end crap gtx 1650/RX 6400 we else got for slot powered gpu. Also with electricity prices these days, some people would appreciate a card like this if pricing was not mentally high.

It´s no different with my RTX A2000. Boost stock is 1200 MHz but can go to 1350 MHz. With overclock it boost to between 1350 and 1600 MHz depending on game and load on gpu.

 

[Lew Zealand]

This is a weird card.
More CUDAs than a 4070, nearly twice the vram and performance like a 3070....
What the hell...
....and all that so they keep it below 70W....

That is the 70W taskmaster of restricting to slot-power only. Don't need that restriction? There are better value cards in the Ada range with PSU power connectors.

 

[Solaris17]

However i am sure many would love a card like this, if the price was a lot lower.

I would be, tempted to do this instead of an A2000 these models wont train themselves.

I really need to get off my K1200

 

[Gucky]

Hmm a cut down AD104 with sandwiched memory and low power...
Why is the price sooo high.
It would be an awesome SFF Card...

 

[TheinsanegamerN]

Interesting card, but priced higher than a rtx 4080, while only perform like a rtx 3070. Not worth it.

I am all ready covered with rtx a2000 and a rtx 4090. So I'll pass on this one.

However a card with 8-10 gb vram and a lot cheaper. I could be interested. For now it's prised to high.

Perf like a 3070 would be fantastic for SFF builds. I look forward to buying one of these when they hit ebay in 2 years for $250 like the A2000 is now.

 

[Tomgang]

Perf like a 3070 would be fantastic for SFF builds. I look forward to buying one of these when they hit ebay in 2 years for $250 like the A2000 is now.

Correct that would be nice. However it cost more than a rtx 4080.

Rtx a2000 performance like a rtx 3050 at times close to rtx 3060, so it is really not that much slower than rtx 4000.

Rtx a2000 msrp was around 450 usd for the 6 gb vram variant while this is more like over 1200 usd. Rtx a2000 gave better performance for it price than a4000. Of cause the 20 GB vram is one of tre reasons why it cost 3 times more, but still, it one expensive rtx 3070 with 20vgb vram.

 

[playerlorenzo]

Sadly no cheaper consumer versions of these will be made.

 

[tvshacker]

so, 20GB of VRAM with 160 bit bus, but they "can't" manage to launch a 4070 with 16GB with 192bit interface???

(or a 4060TI...)

 

[gffermari]

.... but still, it one expensive rtx 3070 with 20vgb vram.

...or a 2080Ti with a truck load of vram.

Spoiler

oh wait.

Anyway...the 70W is impressive though for a 2080Ti/3070 level of performance.

 

[Minus Infinity]

$1000 basically for validated drivers!

 

[lexluthermiester]

I really need to get off my K1200

Really? I thought you'd have been on some much newer...

 

[Solaris17]

Really? I thought you'd have been on some much newer...

Not yet, that machine just trains models all day.

 

[Metroid]

hahahahahahahah 1444 usd for a 4060 sfx version.

 

[oxrufiioxo]

hahahahahahahah 1444 usd for a 4060 sfx version.

This probably what the 4060 would have been if Nvidia felt any pressure from AMD to release decent cards below 1000 usd......

 

[Redwoodz]

This probably what the 4060 would have been if Nvidia felt any pressure from AMD to release decent cards below 1000 usd......

Are you kidding me? Blame someone else for Nvidia's greed? You people are lost. You deserve a $2,000 "RTX" machine that can play lightshows at 12fps.

 

[rattlehead99]

so, 20GB of VRAM with 160 bit bus, but they "can't" manage to launch a 4070 with 16GB with 192bit interface???

(or a 4060TI...)

The bandwidth is halfved on the RTX A4000 ADA. Only 280GB/s compared to the 540GB/s of the RTX 4070(Ti). With a 192 bit bus, you can have up to 6x 32 bit GDDR6(X) chips at a time and GDDR6(X) caps out at 2GB per chip which means it caps out at 12GB. If you want to double it, you'd have to cut the bandwidth in half. With a 160 bit bus you can have 5x 32 bit GDDR6(X) chips or 10x chips GDDR6(X) running at 16 bit each.

 

[rattlehead99]

This is a weird card.
More CUDAs than a 4070, nearly twice the vram and performance like a 3070....
What the hell...
....and all that so they keep it below 70W....

The bandwidth is cut in half and so are the clock speeds. With a 160 bit bus you can have 5x 32 bit GDDR6(X) chips and they cap out at 2GB per chip, thus 10GB maximum at full bandwidth, while with a 192 bit bus you can do 6x 32 bit chips at full bandwidth, and thus 12GB VRAM. If you put more, in this case double(10x chips) you have to run them at 16 bit mode and the bandwidth is cut in half. The RTX A4000 ADA has 280GB/s bandwidth while the 4070(Ti) has 540GB/s.

 

[tvshacker]

The bandwidth is halfved on the RTX A4000 ADA. Only 280GB/s compared to the 540GB/s of the RTX 4070(Ti). With a 192 bit bus, you can have up to 6x 32 bit GDDR6(X) chips at a time and GDDR6(X) caps out at 2GB per chip which means it caps out at 12GB. If you want to double it, you'd have to cut the bandwidth in half. With a 160 bit bus you can have 5x 32 bit GDDR6(X) chips or 10x chips GDDR6(X) running at 16 bit each.

I assumed the bandwidth reduction (vs the 4070) was just because of lower bus and slower chips (gddr6 vs gddr6x on the 4070).
The 4060TI has an estimated 288GB/s on a 128 bit bus so that makes believe the bus is not halved on this 20gb card and the main difference is just the speed of the chips (since they're going for low power).