MLID
I think it's obvious why Nvidia stopped production on the 40 series months ago. It wasn't to ramp up production of the 50 series. It was to exhaust the supply of 40 series so that Nvidia wouldn't be competing with itself.
Sure it can. IIRC RT is more memory hungry so it benefits more from larger cache and memory bandwidth vs rasterization. 5070 is a smaller, stripped down die than even the 4070. It’s safe to assume its memory system has been chopped down.Not Trust.
What's the source of that nonsense?
It makes ZERO sense that 5070 Ti would be relatively faster in RT, but the 5070 would relatively much worse at RT.
Sure it can. IIRC RT is more memory hungry so it benefits more from larger cache and memory bandwidth vs rasterization. 5070 is a smaller, stripped down die than even the 4070. It’s safe to assume its memory system has been chopped down.
No it isn't. 5070 has more cache, and a LOT more memory bandwidth (672 vs 504) over the 4070. It's RT improvements should be similar to raster improvements, or better.
You're assuming the RTX 4070's RT performance was bandwidth-limited, whereas it's probably memory-limited. And if the 5070 has the same amount of VRAM, that problem isn't going away.
It's didn't help to 5080 vs 4080, despite 4080 have 76 RT cores, 5080 84 and higher memory bandwidth"RT is more memory hungry so it benefits more from larger cache and memory bandwidth
It's didn't help to 5080 vs 4080, despite 4080 have 76 RT cores, 5080 84 and higher memory bandwidth
11% only
4070 have 46, 5070 have 48
That was just answer, not to youYou should direct you reply to him, as that's his quote.
The first lesson for 5090 owners: when connectors are the more likely weak link, better wire gauge does little to improve the situation.We have the first known 5090 melting incident. Glad to see things won't change.
Occurred on the PSU side. He notes the PSU 12vhpwr pins are thinner than the ones on the GPU side.
We have the first known 5090 melting incident. Glad to see things won't change.
It melted on both ends. I guess you're saying the initial failure was PSU side?Occurred on the PSU. He notes the PSU 12vhpwr pins are thinner than the ones on the GPU side.
Yes, and there are two things to consider here:It melted on both ends. I guess you're saying the initial failure was PSU side?
The third party cable says its 12v2x6 spec, but the page is kind of ambiguous using 12vhpwr and 12v2x6 seemingly interchangeablyYes, and there are two things to consider here:
The owner claims their PSU is Asus Loki SFX-L, and had it for 2 years now. The new connector is only a bit over 1 year old. On the product page for the Loki, Asus talks about ATX 3.0 compatibility. Based on all of this, I think the connector is 12HPWR.
- one subjective observation that the PSU connector has thinner pins (could be relevant, but it's hard to evaluate from our PoV)
- one bigger question on whether the PSU conector is 12HPWR or 12V-2x6
If this is true, then we get a second moral of the story - if your PSU has a 12HPWR connector, the new 12V-2x6 cable and safer specs won't help. One must avoid the PCB 12HPWR connector like the plague.
The third party cable says its 12v2x6 spec, but the page is kind of ambiguous using 12vhpwr and 12v2x6 seemingly interchangeably
The cable yes, but the big changes made by the new spec are to the PCB connector (on the card and the PSU). The cable side specs are almost identical, albeit with more detailed design criteria (both electrical and mechanical).The third party cable says its 12v2x6 spec
the cable - https://www.moddiy.com/products/ATX...-16-Pin-to-16-Pin-PCIE-Gen-5-Power-Cable.htmlWho made the third party cable?
Buildzoid already posted a rant video...
tl;dw - appears the aftermarket cable implemented the 12v2x6 spec so the shorter sense pins may not be enough to prevent burnout.
Buildzoid already posted a rant video...
tl;dw - appears the aftermarket cable implemented the 12v2x6 spec so the shorter sense pins may not be enough to prevent burnout.