Due to time constraints and because everything has already been reported in advance on YouTube, I only picked out one game, but it offers DLSS, DLSS3 and FSR, so it offers comparisons between DLSS and FSR, as well as DLSS with and without frame generation. i’m so fair to still bring this, even if it took some effort. As always, the bars speak for themselves, and I’ll get to the subject of latencies separately in a moment. Let’s look at the FPS numbers first. And oops, look at the min FPS and compare between DLSS and DLSS 3.0!
So now let’s take a closer look at both comparisons as curve diagrams:
Percentiles in the course
The individual percentiles are also highly interesting, because it covers even more than just the P1 Low:
Frame Times as percentile bars
Variances in direct comparison
Power consumption of the single cards in the benchmark
Efficiency of the individual cards in the benchmark
And finally, here are two more links about DLSS 3.0:
Power consumption in factory state as summary
The 52 or so watts at idle is a telemetry and driver air number, but I already wrote a preliminary note about that on page one. I already communicated the entire problem of power consumption under load, i.e. the significantly higher values in gaming (up to 133 watts e.g. in Cyberpunk 2077 or Horizon Zero Dawn) with the non-OC card as well as the insanely high telemetry values in idle, to NVIDIA last (!) week, but I still haven’t received a satisfactory answer.
By the way, we can see very nicely that the full TBP of the released 121 watts is almost always exceeded. Even the measured values of NVIDIA’s PCAT and Powenetics V2 confirm my results and clearly refute NVIDIA’s own values via the NVAPI interface. Whoever came up with this should have known that something like this would stand out. But it fits well to the PR slide with the funny power saving plan.
Load distribution between PCIe slot and the PCIe sockets
NVIDIA uses the PCIe slot a bit, especially for gaming. But you’re still far below what’s allowed with 5.5 amps maximum for the flowing currents, even with the increased power limit.
Rough 20-ms intervals
Let’s first take a look at the flowing currents. Measurements were taken in coarser 20-ms intervals, i.e. around 50 times per second, to simulate the load on the supervisor chip of the power supplies (shutdown). We see that even short load peaks are capped at 13 A at the latest. but a real limitation looks different.
Nevertheless, we still have to take a look at the voltages, or the product of voltage and current flow. I already wrote that I measured here at different power supply connections, even if in the end on the board of the graphics card all connections meet again somehow. What we can now see here as much clearer fluctuations and peaks is due to the partially a little bit overvolting power supply and thus the voltage and not the currents. This is due to technical reasons, but it’s not a big deal. However, we can also see that the few peaks at 200 watts are not caused by the flowing current (graphics card), but actually result from the power supply! This is also due to the topology and especially the rather puny input filter with only one coil on the single rail.
The torture test is hardly any different.
If you now add the voltage again, then you see a stronger ripple, which again results from the somewhat jittery operating voltage. However, to save the power supply’s honor, it has to be said that this affects all current products from all manufacturers and can hardly be avoided.
- 1 - Einführung, technische Daten und Technologie
- 2 - Test System im igor'sLAB MIFCOM-PC
- 3 - Teardown: PCB und Komponenten
- 4 - Gaming Performance FHD (1920 x 1080)
- 5 - Gaming-Performance WQHD (2560 x 1440)
- 6 - Gaming Performance DLSS vs. DLSS3 vs. FSR
- 7 - Lastspitzen, Kappung und Netzteilempfehlung
- 8 - Temperaturen, Taktraten, Lüfter und Geräuschentwicklung
- 9 - Zusammenfassung und Fazit
20 Antworten
Kommentar
Lade neue Kommentare
Urgestein
1
Urgestein
Veteran
Mitglied
Urgestein
Veteran
Veteran
Veteran
Urgestein
Urgestein
Veteran
Mitglied
1
Mitglied
Moderator
Neuling
Neuling
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →