GPU leaker @Kopite7kimi (opens in new tab) is at it once more immediately with one other replace on the upcoming RTX 4080 specs. His newest rumors embrace a core depend discount for the RTX 4080 from the estimated 10,240 to 10,752 vary all the way down to 9728. This interprets to a ten% discount in core depend.
Fortunately, this transformation would not seem to alter the 4080’s estimated Time Spy Excessive rating of 15,000, which Kopite lists once more in his newest Tweet. Efficiency seems to stay the identical, for now no less than. Kopite’s different indicated specs for the RTX 4080 additionally stay intact, together with the 420W energy restrict, 16GB of 21Gbps GDDR6X reminiscence, and AD103 die.
Whatever the supply, apply a wholesome dose of skepticism to those rumors and “leaks.” Errors and incorrect guesses are inevitable, and the {hardware} specs may not be totally nailed down but. One crucial lacking piece of data are the RTX 40-series core clocks. There’s hypothesis, primarily based on the upper energy ranges, that Nvidia will enhance core clocks considerably relative to the prevailing Ampere line, however there isn’t any exhausting information as but.
I am not a chatterbox, however I’ve to make some updates. I hope you do not thoughts. a potential RTX 4080, PG136/139-SKU360AD103-300-A19728FP32256bit 16G 21Gbps GDDR6Xtotal energy ~420WTSE ~15000Now I’ve accomplished the most recent replace for 4090, 4080 and 4070.August 5, 2022
RTX 40-Sequence Titan Rumor and L2 Cache Updates
One other Tweet from @kopite7kimi suggests the Titan branding could also be coming again to the GeForce lineup in some form or type. This RTX 40-Sequence Titan “has an opportunity” at utilizing the full-fat AD102 chip, however Kopite hasn’t confirmed whether or not or not that can truly occur. Extra probably is it is going to be a virtually full AD102, since 100% practical chip yields on the presumably massive GPU will probably be low.
There’s an fascinating query if we do in truth get a brand new Titan card. Will Nvidia as soon as once more have a Titan “hybrid” SKU that blurs the traces between Nvidia’s A-series workstation GPUs and the corporate’s GeForce gaming GPU lineup, with a lot of reminiscence and enhanced drivers? Or will we as a substitute get an RTX 4090 Ti that skips the skilled driver enhancements? Both approach, we are able to safely guess a card with 48GB of GDDR6X reminiscence will not come low-cost. Kopite7kimi seems to have supplied an informed guess for now could be all.
In the identical Tweet, Kopite additionally shared some L2 cache updates for the AD103 and AD104 40 sequence dies. The totally unlocked AD103 die will apparently have 64MB of L2 cache, whereas the AD104 die may have 48MB — that is the identical as what the sooner leaks steered again initially of the yr.
Nevertheless, Kopite additionally suggests the RTX 4090 might need a diminished L2 cache dimension, or maybe that AD102 will not have proportionately bigger caches. Or it could be a transfer to distinguish the potential RTX 4090 Ti (or Titan) from the “lesser” RTX 4090.
1. The TITAN have an opportunity to make use of a full-fat chip. I’ll replace that once I verify it. 2. A full-fat AD103 may have 64M L2, and AD104 may have 48M. Solely RTX 4090 ought to have a L2 reduce. (99%😁)August 5, 2022
