We return to the fray with new rumors and these indicate that the NVIDIA Geforce RTX 4080 will have a TGP of 420W. That would mean 100W more than its predecessor, the RTX 3080, which had a value of 320W. This would be another case of increasing power consumption in the footsteps of the RTX 4060. A single day difference between the two clearly shows that the RTX 4070 will most likely end up consuming more.
The RTX 4080 will have a consumption of 420W, but a low-end chip
Possible RTX 4080, PG139-SKU360, has 420W TGP, still uses AD103.
– kopite7kimi (@kopite7kimi) June 6, 2022
These new rumors come from one of our dear leakers, @kopite7kimi, who recently mentioned the TGP of the Geforce RTX 4080. On top of that, he comments that it will continue to use the chip AD103 TSMC’s 5nm Lovelace architecture. Now, for the most attentive readers, we imagine that you noticed a detail and is that it mentions TGP and not TDP.
This means that this time it refers to the total consumption of the graph itself, because TGP is “Full graphics power”. On the other hand, we have the acronym PDT which stand for “Thermal Design Power”, which indicates the consumption only of the core of the GPU. Therein lies the interesting thing, is that the TGP will always be upper to the TDP, since the latter does not indicate the total consumption.
For this reason, when we go to see reviews of graphics cards and proceed to observe the consumption in the benchmarks, this is usually very high. In most cases much higher than what the manufacturer originally advertised. This is due to the fact that generally the measure they offer us is that of the PDTbeing a value not representative of total consumption.
This measurement remains quite stable and although we do not have more recent data, we can assume that the difference in consumption will be similar. Keep in mind that the TGP adds the rest of the components outside of the core graph, such as the VRAM. The problem is that NVIDIA, contrary to what many think, actually offers TGP as a measurement today and since the RTX 20, but many still think it’s TDP.
This creates a problem which also adds to the fact revealed by Kopite about the chip used.
The RTX 4080 won’t feature Ada Lovelace’s best chip
This is perhaps the most disturbing thing about it. The RTX 4080 will feature an AD103 chip with a higher or better midrange SKU, which it will surely share with the RTX 4070. consumption, where those 420 watts are just the starting point.
Considering that the RTX 4060’s alleged AD104 will consume 220 watts and there are two theoretical models above it, it’s more than likely that 600 watts per GPU is something NVIDIA doesn’t want to support because it’s more than what it is ready to offer. when the RTX 40 was born. We already have the first data, we just have to start realizing that it is more than likely that a huge number of users will have to change their power supply if they want to do with one of these models, while manufacturers rub their hands…
Via: video card