By: Domaldel (moris.b.rubeksson.delete@this.gmail.com), January 12, 2019 1:20 am
Room: Moderated Discussions
noone (noone.delete@this.noone.com) on January 11, 2019 9:05 pm wrote:
> Domaldel (moris.b.rubeksson.delete@this.gmail.com) on January 11, 2019 5:37 pm wrote:
> > Jeff S. (fakity.delete@this.fake.com) on January 10, 2019 1:53 pm wrote:
> > > someguy (someguy.delete@this.gmail.com) on January 10, 2019 1:06 pm wrote:
> > > > Any comments on AMD's announcements at CES?
> > >
> > > Disappointingly little was actually covered:
> > > - They are rebranding MI50 HPC cards (16 GiB HBM2 and 60 CUs) as $700 gaming
> > > card and are in all likelihood not hoping/planning to sell all that many.
> > > - Navi was not even announced and hence unlikely to show up before the Holidays in the best case.
> > > - An ES (or QS?) of the 8c Ryzen can tie i9-9900k in SSE2-based CB15 and maybe a game or two.
> > > - There was a conspicuous empty space on the substrate that is virtually
> > > a confirmation of 12c-16c parts with a 2nd compute chiplet.
> > > - Expect actual product launches in Summer.
> > >
> > > The biggest takeaway I had was that Zen 2 is not outright
> > > crippled by off-die memory controllers or the rumored
> > > poor TSMC 7nm clocking, at least for a couple cherrypicked tests. Most current speculative debates seem to
> > > circle around if/how badly 16c AM4 chips will be bottlenecked
> > > by 2 DDR4 channels, but this seems hugely less
> > > interesting than all the unanswered questions around design and performance changes in the core.
> > >
> > > The lack of any new hardware from RTG is AMD's bigger issue at the moment I feel.
> >
> > Correct me if I'm wrong, but doesn't the Radeon VII have significantly higher number of ROPs then the MI50?
> >
> > Yes, MI50 only has 64 ROPs while Radeon VII has 128.
> > On the other hand Radeon VII also has fewer compute units and no error correction.
> > And we don't know if it'll have half rate DP performance or not yet.
> > If it does I might seriously consider getting it.
>
> Nope. It's the same (64). Some media outlets misreported.
Crap...
In that case I'm a bit less enthusiastic about this thing...
> Domaldel (moris.b.rubeksson.delete@this.gmail.com) on January 11, 2019 5:37 pm wrote:
> > Jeff S. (fakity.delete@this.fake.com) on January 10, 2019 1:53 pm wrote:
> > > someguy (someguy.delete@this.gmail.com) on January 10, 2019 1:06 pm wrote:
> > > > Any comments on AMD's announcements at CES?
> > >
> > > Disappointingly little was actually covered:
> > > - They are rebranding MI50 HPC cards (16 GiB HBM2 and 60 CUs) as $700 gaming
> > > card and are in all likelihood not hoping/planning to sell all that many.
> > > - Navi was not even announced and hence unlikely to show up before the Holidays in the best case.
> > > - An ES (or QS?) of the 8c Ryzen can tie i9-9900k in SSE2-based CB15 and maybe a game or two.
> > > - There was a conspicuous empty space on the substrate that is virtually
> > > a confirmation of 12c-16c parts with a 2nd compute chiplet.
> > > - Expect actual product launches in Summer.
> > >
> > > The biggest takeaway I had was that Zen 2 is not outright
> > > crippled by off-die memory controllers or the rumored
> > > poor TSMC 7nm clocking, at least for a couple cherrypicked tests. Most current speculative debates seem to
> > > circle around if/how badly 16c AM4 chips will be bottlenecked
> > > by 2 DDR4 channels, but this seems hugely less
> > > interesting than all the unanswered questions around design and performance changes in the core.
> > >
> > > The lack of any new hardware from RTG is AMD's bigger issue at the moment I feel.
> >
> > Correct me if I'm wrong, but doesn't the Radeon VII have significantly higher number of ROPs then the MI50?
> >
> > Yes, MI50 only has 64 ROPs while Radeon VII has 128.
> > On the other hand Radeon VII also has fewer compute units and no error correction.
> > And we don't know if it'll have half rate DP performance or not yet.
> > If it does I might seriously consider getting it.
>
> Nope. It's the same (64). Some media outlets misreported.
Crap...
In that case I'm a bit less enthusiastic about this thing...
| Topic | Posted By | Date |
|---|---|---|
| AMD CES | someguy | 2019/01/10 02:06 PM |
| AMD CES | Jeff S. | 2019/01/10 02:53 PM |
| AMD CES | Domaldel | 2019/01/11 06:37 PM |
| AMD CES | noone | 2019/01/11 10:05 PM |
| AMD CES | Domaldel | 2019/01/12 01:20 AM |
| AMD CES | arifureta | 2019/01/12 07:09 AM |
| AMD CES | Domaldel | 2019/01/13 01:57 AM |
| AMD CES | Montaray Jack | 2019/01/17 10:00 AM |
| AMD CES | Montaray Jack | 2019/01/17 12:32 PM |
| AMD CES | hobold | 2019/01/10 03:17 PM |
| AMD CES | Jan Olšan | 2019/01/10 03:30 PM |
| AMD CES | Adrian | 2019/01/10 09:37 PM |
| AMD CES | Alberto | 2019/01/11 11:13 AM |
| AMD CES | JS | 2019/01/12 02:05 AM |
| AMD IO die | anon³ | 2019/01/13 02:17 PM |
| AMD IO die | anon | 2019/01/13 03:26 PM |
| eDRAM too expensive | David Kanter | 2019/01/13 03:34 PM |
| eDRAM too expensive | Adrian | 2019/01/13 09:51 PM |
| eDRAM too expensive | Adrian | 2019/01/13 10:01 PM |
| eDRAM too expensive | Ricardo B | 2019/01/14 12:06 PM |
| Zen PHY speeds | Jeff S. | 2019/01/14 12:40 PM |
| eDRAM too expensive | Adrian | 2019/01/14 01:44 PM |
| Zen target process (was: eDRAM too expensive) | hobold | 2019/01/14 03:42 PM |
| Zen target process (was: eDRAM too expensive) | Adrian | 2019/01/14 10:13 PM |
| Zen target process (was: eDRAM too expensive) | hobold | 2019/01/15 09:28 PM |
| eDRAM too expensive | Ricardo B | 2019/01/15 04:40 AM |
| eDRAM too expensive | Ricardo B | 2019/01/15 09:26 AM |
| eDRAM too expensive | Maynard Handley | 2019/01/14 11:38 AM |
| eDRAM too expensive | anon | 2019/01/14 02:00 PM |
| eDRAM too expensive | Maynard Handley | 2019/01/14 02:43 PM |
| eDRAM too expensive | Dan Fay | 2019/01/14 03:16 PM |
| eDRAM too expensive | Maynard Handley | 2019/01/14 04:40 PM |
| eDRAM too expensive | Dan Fay | 2019/01/14 05:58 PM |
| eDRAM too expensive | someone | 2019/01/15 09:20 AM |
| eDRAM too expensive | anon | 2019/01/14 04:34 PM |
| eDRAM too expensive | Maynard Handley | 2019/01/14 06:34 PM |
| eDRAM too expensive | anon | 2019/01/15 02:06 AM |
| AMD IO die | juanrga | 2019/01/14 04:04 AM |
| AMD IO die | Maynard Handley | 2019/01/14 11:42 AM |
| AMD IO die | Alberto | 2019/01/14 03:00 PM |
| AMD IO die | juanrga | 2019/01/15 03:54 AM |
| AMD IO die size & functionality | Wes Felter | 2019/01/15 12:12 PM |
| AMD IO die size & functionality | juanrga | 2019/01/15 12:43 PM |
| AMD IO die size & functionality | Alberto | 2019/01/15 01:57 PM |
| AMD CPU die size & functionality | Wes Felter | 2019/01/15 05:35 PM |
| AMD CPU die size & functionality | anon | 2019/01/16 02:32 AM |
| AMD CPU die size & functionality | Alberto | 2019/01/16 06:40 AM |
| AMD CPU die size & functionality | Jan Olšan | 2019/01/16 08:11 AM |
| AMD IO die size & functionality | Maynard Handley | 2019/01/16 10:17 AM |