Activity Stream

Filter
Sort By Time Show
Recent Recent Popular Popular Anytime Anytime Last 7 Days Last 7 Days Last 30 Days Last 30 Days All All Photos Photos Forum Forums
  • psoomah's Avatar
    21st July 2018, 04:44 PM
    psoomah replied to a thread Epyc 2 in CPUs
    Maybe if these were all being fabricated on 14nm. Unlikely to be feasible on 7nm.
    51 replies | 4000 view(s)
  • testbug00's Avatar
    21st July 2018, 04:22 PM
    testbug00 replied to a thread Epyc 2 in CPUs
    If it hooks up to each chiplet of cores, where else would you put it but the middle? Like, I see other ways to do it (can mock up in paint once...
    51 replies | 4000 view(s)
  • psoomah's Avatar
    21st July 2018, 04:10 PM
    psoomah replied to a thread Epyc 2 in CPUs
    Considering the order of magnitude increase in complexity and expense of putting one 7nm CPU design into production it is highly unlikely AMD would...
    51 replies | 4000 view(s)
  • mrzombiewithbrains's Avatar
    21st July 2018, 04:09 PM
    mrzombiewithbrains replied to a thread Epyc 2 in CPUs
    It's not middle, it's main chip. And it's not positioned in the middle either, it would be totally illogical to do that. Don't look at adoredtv...
    51 replies | 4000 view(s)
  • testbug00's Avatar
    21st July 2018, 04:05 PM
    testbug00 replied to a thread AMD Navi on 7nm Information in GPUs
    Vegais efficient if you run it at low clocks/bin it. Vega 64/56 are not binned. In fact downvolted/downclocked Vega64 at a higher clockspeed...
    909 replies | 111847 view(s)
  • mrzombiewithbrains's Avatar
    21st July 2018, 03:53 PM
    mrzombiewithbrains replied to a thread Epyc 2 in CPUs
    Of course they are, they would not waste time in doing the exact same chip on both foundries. They announced that zen2 design was completed many...
    51 replies | 4000 view(s)
  • Stuckey's Avatar
    21st July 2018, 03:30 PM
    Stuckey replied to a thread Epyc 2 in CPUs
    Desktop 7nm Ryzen could be 12C/24T, I think that's entirely possible. As I just pointed out, Zeppelin wastes quite a lot of area by having IF links...
    51 replies | 4000 view(s)
  • mrzombiewithbrains's Avatar
    21st July 2018, 03:20 PM
    The zen L2 is shared between all it's cores, but it's not shared in summit ridge because the cross talk CCX IF would have given it an huge latency...
    2335 replies | 410608 view(s)
  • Kedas's Avatar
    21st July 2018, 03:14 PM
    Kedas replied to a thread Epyc 2 in CPUs
    The leak confusion is probably caused due to the different CPU designs of EPYC and TR and AM4 dies. So they will probably all be true, 48C and 64C....
    51 replies | 4000 view(s)
  • mrzombiewithbrains's Avatar
    21st July 2018, 02:55 PM
    Now suddenly vega is efficient, come guys I have been telling to not look at vega64, take a look at vega56 or some hypothetical vega 48 if existed or...
    909 replies | 111847 view(s)
  • mrzombiewithbrains's Avatar
    21st July 2018, 02:34 PM
    Interesting you got the same math I did when I published the wccftech link where they say 70% higher IPC vs vega10. 12TF + 70% = 20.4TF ;)
    909 replies | 111847 view(s)
  • Zizy's Avatar
    21st July 2018, 01:44 PM
    Eh, he is probably right that 5% IPC is more realistic than 10%+. 5% clocks is humhum. If we see 12C instead of 8C this should be about right if...
    2335 replies | 410608 view(s)
  • Zizy's Avatar
    21st July 2018, 01:37 PM
    Zizy replied to a thread Epyc 2 in CPUs
    To get from 48C to 64C, there are 2 options. 1: 48C was 6*8C and now it is 8*8C. I find this incredibly implausible - that AMD planned 6 chips on...
    51 replies | 4000 view(s)
  • french toast's Avatar
    21st July 2018, 12:45 PM
    As mentioned earlier/different thread, I don't doubt EUV and 7nm brings cost benefits in the long term, I have serious doubts whether we see any EUV...
    2335 replies | 410608 view(s)
  • Seronx's Avatar
    21st July 2018, 09:19 AM
    EUV reduces costs by ~30% and increases yields by ~50%. Overall, AMD will be near 28-nm/20-nm yields. So, with that info on 7Hybrid-EUV they can go...
    2335 replies | 410608 view(s)
  • french toast's Avatar
    21st July 2018, 08:43 AM
    Cheers, but are we not taking into account yeilds on expensive 7nm process? I would think they would want a much smaller die, how much would all the...
    2335 replies | 410608 view(s)
  • Seronx's Avatar
    21st July 2018, 08:11 AM
    The info mostly comes from the slides. Speculation: TSMC and GlobalFoundries have the near-exact mask count and density. -> Logic is 6-track...
    2335 replies | 410608 view(s)
  • psoomah's Avatar
    21st July 2018, 07:51 AM
    psoomah replied to a thread Epyc 2 in CPUs
    Translating available 7nm manufacturing capacity, the single biggest barrier to AMD's potential market growth, into the maximum possible number of...
    51 replies | 4000 view(s)
  • french toast's Avatar
    21st July 2018, 05:28 AM
    Ha ha, you are starting to crack me up...I started this whole 'be realistic' thing, you have taken it to the next level!, By this rate zen 3 on 7nm+...
    2335 replies | 410608 view(s)
  • french toast's Avatar
    21st July 2018, 03:45 AM
    Yea I think Navi will be a gaming card, midrange.
    909 replies | 111847 view(s)
  • Stuckey's Avatar
    21st July 2018, 02:18 AM
    Stuckey replied to a thread Epyc 2 in CPUs
    Middle control die has how many memory channels/IF links and associated I/O pads ? Eight of each ? (Six of which would be pretty much useless on AM4...
    51 replies | 4000 view(s)
  • Ragnar's Avatar
    21st July 2018, 01:44 AM
    Well maybe this is what is boils down to, i can also not say i have tried EVERY android and every windows smartphone, so my opinion is based on two...
    40 replies | 1963 view(s)
  • testbug00's Avatar
    21st July 2018, 01:13 AM
    testbug00 replied to a thread Epyc 2 in CPUs
    Actually, if it is 5 die, instead of producing being a uncore die; to allow reuse we could see: 1. Middle Control die with 6/8/12 cores (reused for...
    51 replies | 4000 view(s)
  • Stuckey's Avatar
    21st July 2018, 01:00 AM
    Stuckey replied to a thread AMD Navi on 7nm Information in GPUs
    Wasn't there a rumor a few months back about a '<$300 card with GTX1080 performance' ? My initial thought on seeing that was 'Sounds like the...
    909 replies | 111847 view(s)
  • Stuckey's Avatar
    21st July 2018, 12:24 AM
    Stuckey replied to a thread Epyc 2 in CPUs
    According to Charlie there was a 48C 'Starship' on the roadmap, but that got ditched for something even more monstrous. So best guess would be 64C.
    51 replies | 4000 view(s)
  • Nerdmaster's Avatar
    20th July 2018, 10:26 PM
    Amd would be lucky if they achieve >5% ipc gain compared to 2xxx ryzen or >10% compared to 1xxx ryzen. This gain will already bring them on par with...
    2335 replies | 410608 view(s)
  • Seronx's Avatar
    20th July 2018, 10:08 PM
    Based on Zen slides, the 15% IPC is an under-estimate. Zen x+1 has been disclosed by AMD to be ~50% above x mostly always. At AMD, their 7LP is...
    2335 replies | 410608 view(s)
  • testbug00's Avatar
    20th July 2018, 09:45 PM
    testbug00 replied to a thread AMD Navi on 7nm Information in GPUs
    i can't see that. 2H2019, 1H2020 if late...
    909 replies | 111847 view(s)
  • testbug00's Avatar
    20th July 2018, 09:45 PM
    To keep products launches and content about them spread out. Stay in the news.
    124 replies | 12020 view(s)
  • chithanh's Avatar
    20th July 2018, 08:59 PM
    Even disregarding that it would probably be an unwise move to launch the 16 core ahead of the 24 and 32 core TR2000: Ryzen 2000 launched almost...
    124 replies | 12020 view(s)
More Activity
WordPress Appliance - Powered by TurnKey Linux