Home » today » Technology » Nvidia quietly released the GeForce GT 1010 or Pascal to the bone

Nvidia quietly released the GeForce GT 1010 or Pascal to the bone

If you feel déjà vu, believe that it is only apparent. This is the year we first heard about the GeForce GT 1010. At the beginning of January 2021, there was a mention of this card in the official drivers from Nvidia, and if we can trust the data from the TechPowerUp website, the card was1 issued. A year later, however, there was neither sight nor hearing, only now does it begin to appear in the first photographs.

At first, it was in the images shown above in the form of a semi-generic green card, to which no brand applies (PCBs and stickers do not bear the logo of any manufacturer), and then photos of the retail version from Leadtek appeared:

What exactly is the GeForce GT 1010: It is basically a GeForce GT 1030 (core GP108), which has been further trimmed: The number of stream processors and texturing units has dropped by a third (384-> 256, 24-> 16) and the number of ROPs units dropped straight to half (16-> 8). The card is equipped with a 64-bit bus that can be fitted with 2 GB of memory, either (effectively) 6GHz GDDR5 or 2.1GHz DDR4. The generation 3.0 PCIe interface is present on four lines.

If you’re trying to get an idea of ​​the graphics performance of this product, it might tell you that the already full-featured GeForce GT 1030 can handle current integrated graphics quite successfully (and there’s no need to consider the new APU Rembrandt). The 33-50% slow-down version, which may still have 65% reduced memory throughput (in the case of DDR4 memory), will offer a fraction of the performance of last year’s / previous year’s integrated graphics from AMD or Intel mobile processors.

GeForce
GT 1010
GeForce
GT 1030
GeForce
GTX 1050
DDR4 GDDR5 DDR4 GDDR5
GPU GP108
1.8 billion tr.
GP107
3.3 billion tr.
Area 70 mm² 135 mm²
Process 14 nm Samsung 14 nm Samsung
Architect. Pascal
Rate 1228 MHz 1151 MHz 1228 MHz 1354 MHz
Boost 1468 MHz 1379 MHz 1468 MHz 1455 MHz
SPs 256 384 640
TMUs 16 24 40
ROPs 8 16 32
FP32 0,75 TFLOPS 1,06 TFLOPS 1,13 TFLOPS 1,86 TFLOPS
FP64 0,02 TFLOPS 0,03 TFLOPS 0,03 TFLOPS 0,05 TFLOPS
Memoirs 2 GB
64bit
DDR4
2 GB
64bit
GDDR5
2 GB
64bit
DDR4
2 GB
64bit
GDDR5
2 GB
128bit
GDDR5
Rate 2,1 GHz 6 GHz 2,1 GHz 6 GHz 7 GHz
That. prop. 16,8 GB/s 48 GB/s 16,8 GB/s 48 GB/s 112 GB/s
consumed 3D 30 W 30 W 20 W 30 W 75 W
Power supply
Outputs HDMI + D-Sub
/
2× HDMI
according to the manufacturer DVI
HDMI 2.0b
DVI
DP 1.2/1.4
HDMI 2.0b
Interface PCIe 3.0 ×4 PCIe 3.0 ×16
API DirectX 12
SLI
date 2021 / 2022? April 2018 17. 5. 2017 18. 10. 2016
25. 10. 2016
price ~$70 ? $70 $109

The performance of the GeForce GT 1010 can also be compared with separate graphics cards: In the Time Spy Extreme test, the Radeon RX 6500 XT achieves 2475 points (see e.g. here).

The GeForce GT 1010 scores 179 points, which is about 14 times less (7% of Radeon’s performance).

graph-1

It does not make sense to deal with the segment of individual cards, it can only be compared with integrated cards. The question of why the GeForce GT 1010 is slower than them can be answered by comparing paper parameters. First Ryzen 7 5700G, then GeForce GT 1010: Stream-processors: 512 / 256. Core clock: 1468 MHz / 2000 MHz. GeForce GT 1010 data throughput: 16.8 GB / s (DDR4) and 48 GB / s (GDDR5); for Ryzen 7 5700G 51.2 GB / s (DDR4-3200) and 68.3 GB / s (DDR4-4266). Integrated graphics with system memories therefore have higher data throughput than stand-alone ones equipped with their own memories.

Therefore, the card will obviously not be used anywhere where some 3D performance is expected. However, for a 2D or non-stop running industrial system, it is not entirely ideal to use an active cooler (on both known models), as these small high-speed fans usually have a very limited service life.

The card comes in Asia for (in terms of) $ 70, which is roughly the same price as the GeForce GT 1030 was introduced almost five years ago.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.