|
Well, it makes sense, kinda? If you spent a couple hundred on cpu silicon alone, you're probably getting a separate GPU anyway. If nothing else, it should make cooling more efficient?
|
#
?
Apr 29, 2015 21:46
|
|
|
|
| # ? Jan 25, 2025 06:44 |
|
|
Truga posted:Well, it makes sense, kinda? If you spent a couple hundred on cpu silicon alone, you're probably getting a separate GPU anyway. If nothing else, it should make cooling more efficient? I wonder if it would be possible to just stick in copper plates into the CPU to wick heat away. Usually we get a dead/disabled core, but if we could instead of that have on-chip heatsinks. (I know nothing of this, just musing aloud.)
|
|
#
?
Apr 29, 2015 21:50
|
|
|
No Gravitas posted:I wonder if it would be possible to just stick in copper plates into the CPU to wick heat away. Usually we get a dead/disabled core, but if we could instead of that have on-chip heatsinks. There are already heat shields.
|
|
#
?
Apr 29, 2015 21:54
|
|
|
I must be stupid, has AMD released FX processors this year? Why are they bothering to release any piledriver chips for 2015 at all? What could they possibly offer over anything they currently have that makes it worth the cost? They're not even 28nm?! Am I not reading the roadmap right? Also, wasn't Godavari supposed to be BGA? I guess out of 2015 I'm most excited for Beema, since I've been looking to replace my clunky and oversized old C2Q for HTPC use.
|
|
#
?
Apr 29, 2015 22:25
|
|
|
FaustianQ posted:I guess out of 2015 I'm most excited for Beema, since I've been looking to replace my clunky and oversized old C2Q for HTPC use. In theory, we should be seeing small form factor APU things coming out for exactly that usage. Instead, I'm not seeing anything from AMD so I got one of these. The NUCs are a really impressive tiny-PC package.
|
|
#
?
Apr 29, 2015 22:29
|
|
|
The most interesting part I think is that all socketed desktop chips look like they're returning to a single platform, FM3.
|
|
#
?
Apr 29, 2015 23:06
|
|
|
FaustianQ posted:Also, wasn't Godavari supposed to be BGA?
|
|
#
?
Apr 29, 2015 23:56
|
|
|
FaustianQ posted:I must be stupid, has AMD released FX processors this year? Why are they bothering to release any piledriver chips for 2015 at all? What could they possibly offer over anything they currently have that makes it worth the cost? They're not even 28nm?! Am I not reading the roadmap right? Also, wasn't Godavari supposed to be BGA? Maybe they'll be releasing be a FX-9590E and FX-9370E like they did with the FX-8370(E) and FX-8320(E) last year. I guess enough people are buying their crappy FX series to warrant another revision. The Fitlet is just about the only NUC-class computer I can think of that uses AMD, but it uses Mullins instead of Beema. Actually, I haven't actually seen any Mullins or Beema based computers except the fitlet.
|
|
#
?
Apr 30, 2015 00:01
|
|
|
Rastor posted:You're thinking of Carrizo, in the mobility roadmap. Godavari is the Kaveri with a very slight clockspeed boost. I was sure that Excavator wasn't supposed to have a socketed version though, I must be misremembering. GrizzlyCow posted:Maybe they'll be releasing be a FX-9590E and FX-9370E like they did with the FX-8370(E) and FX-8320(E) last year. I guess enough people are buying their crappy FX series to warrant another revision. I have no idea why, seems like a waste of silicon. Maybe they'll go nuts and release 16 core Opertons as an FX-9950 and 9990 so you can get Zen-like performance out of it. Also, RIP socket AM1 it looks like, I was hoping for K12 on AM1 but it looks like K12 is purely a BGA/Tablet/Mobile part. Maybe FM3 can scale down to AM1, making the whole thing pointless?
|
|
#
?
Apr 30, 2015 00:36
|
|
|
If Zen can give similar performance to Skylake (for similar or less cost) then I'll be very tempted to give AMD a few bucks. The proof will be in the pudding.
|
|
#
?
Apr 30, 2015 15:31
|
|
|
No Gravitas posted:I wonder if it would be possible to just stick in copper plates into the CPU to wick heat away. Usually we get a dead/disabled core, but if we could instead of that have on-chip heatsinks. You don't know where the dead core will be ahead of time. Taking the time and effort to locate one is difficult enough, having a second op to remove that silicon and put something else in would be horrendous for efficiency.
|
|
#
?
Apr 30, 2015 15:45
|
|
|
El Scotch posted:If Zen can give similar performance to Skylake (for similar or less cost) then I'll be very tempted to give AMD a few bucks.
|
|
#
?
Apr 30, 2015 15:48
|
|
|
Boiled Water posted:That is, to say the least, unlikely. He didn't say anything about per-watt performance though 
|
|
#
?
Apr 30, 2015 17:46
|
|
|
roadhead posted:He didn't say anything about per-watt performance though That would be something to see, a 300W Zen locked with a 65W Skylake in multithreaded performance. Also feasible!
|
|
#
?
Apr 30, 2015 18:17
|
|
|
JawnV6 posted:You don't know where the dead core will be ahead of time. Taking the time and effort to locate one is difficult enough, having a second op to remove that silicon and put something else in would be horrendous for efficiency. Oh, I know that. But you could just pick some cores and replace them with metal instead of even attempting fabbing them for cores. You would alternate core metal core metal core. Yes, your yields would suck. But would it actually help to get heat out of the chip? With shrinking processes you could probably just have them in between the cores anyway. I always see the chip crammed on chip photograps, but maybe some metal spacers would help wick the heat out?
|
|
#
?
Apr 30, 2015 18:47
|
|
|
No Gravitas posted:Oh, I know that. In addition to horrible yields you will double the core price per wafer, making a dual-core processor in four-core processor die size. And that would make manufacturing uneconomical.
|
|
#
?
Apr 30, 2015 19:30
|
|
|
Rosoboronexport posted:In addition to horrible yields you will double the core price per wafer, making a dual-core processor in four-core processor die size. And that would make manufacturing uneconomical. Sounds like something AMD would be interested in!
|
|
#
?
Apr 30, 2015 19:40
|
|
|
No Gravitas posted:But you could just pick some cores and replace them with metal instead of even attempting fabbing them for cores. You would alternate core metal core metal core. Yes, your yields would suck. But would it actually help to get heat out of the chip? No Gravitas posted:With shrinking processes you could probably just have them in between the cores anyway. I always see the chip crammed on chip photograps, but maybe some metal spacers would help wick the heat out?
|
|
#
?
Apr 30, 2015 19:43
|
|
|
roadhead posted:He didn't say anything about per-watt performance though Also true. I admit, given I'm using a water cooled setup I'm not particularly concerned about wattage as long as the performance (and price) are good. (you know, as long as it's not 300+ watts on the cpu  )A big part of me really, really wants to be able to unironically buy an AMD chip without excuses. In the end it's up to them to prove they've earned it. Wistful of Dollars fucked around with this message at 22:53 on Apr 30, 2015 |
|
#
?
Apr 30, 2015 22:50
|
|
|
A big part of me wants to be able to buy an AMD chip unironically, too, but my machine needs upgrading now, not later.
|
|
#
?
Apr 30, 2015 23:52
|
|
|
GrizzlyCow posted:Maybe they'll be releasing be a FX-9590E and FX-9370E like they did with the FX-8370(E) and FX-8320(E) last year. I guess enough people are buying their crappy FX series to warrant another revision. These chips may already exist, they bin everything. Who knows how long the 9590 was sitting in inventory before they had enough to sell them.
|
|
#
?
May 1, 2015 17:32
|
|
|
It's AMD’s Financial Analyst Day and that means Zen is official: http://anandtech.com/show/9231/amds-20162017-x86-roadmap-zen-is-in
|
|
#
?
May 6, 2015 18:55
|
|
|
Sidesaddle Cavalry posted:The most interesting part I think is that all socketed desktop chips look like they're returning to a single platform, FM3. In which I got totally wrecked by a fake presentation slide the other day and the single platform is actually called AM4. 40% IPC increase over Excavator. Isn't that only going to bring it up to parity with Haswell?
|
|
#
?
May 6, 2015 20:35
|
|
|
Sidesaddle Cavalry posted:In which I got totally wrecked by a fake presentation slide the other day and the single platform is actually called AM4. Assuming no decrease in clock rates?
|
|
#
?
May 6, 2015 20:55
|
|
|
It's very arbitrary, I know. We don't even have Excavator cores on market yet and I assumed only a very small increase over Steamroller.
|
|
#
?
May 6, 2015 21:21
|
|
|
Haswell performance out of an AMD part is still good, and they seem to acknowledge this since there is a refresh by the end of the year. This also pegs Zen as being 2016 Q1/2, which might mean it gets out before Cannonlake, yes? Also RIP ARM for this generation apparently :\
|
|
#
?
May 6, 2015 21:37
|
|
|
FaustianQ posted:Also RIP ARM for this generation apparently :\ They're still doing ARM, but just their low-power / low-performance Opteron 1100 for 2015 through 2016, and apparently dropping the idea of boards that can accept both ARM and x86 chips (which honestly always sounded goofy to me anyway).
|
|
#
?
May 6, 2015 21:41
|
|
|
Haswell performance is "good enough" for them to be consider a contender if they are priced competitively. Makes me think of the early Athlon and Pentium 3 days. The Pentium 3 was slightly faster in a lot of games because it has stronger FP performance but the Athlon chip was considerably cheaper. A lot people didn't care about 10% performance increase in some applications when you could save 30-40%.
|
|
#
?
May 6, 2015 22:36
|
|
|
Rastor posted:They're still doing ARM, but just their low-power / low-performance Opteron 1100 for 2015 through 2016, and apparently dropping the idea of boards that can accept both ARM and x86 chips (which honestly always sounded goofy to me anyway). I dunno, a K12 desktop sounded like a hilarious waste of money right up my alley, just because I like the idea of a desktop ARM platform.
|
|
#
?
May 6, 2015 22:45
|
|
|
Lowen SoDium posted:Haswell performance is "good enough" for them to be consider a contender if they are priced competitively. Makes me think of the early Athlon and Pentium 3 days. The Pentium 3 was slightly faster in a lot of games because it has stronger FP performance but the Athlon chip was considerably cheaper. A lot people didn't care about 10% performance increase in some applications when you could save 30-40%. Haswell performance with 6/8 core chips for cheaper money could be viable in the DX12/Vulcan world.
|
|
#
?
May 7, 2015 00:39
|
|
|
El Scotch posted:Haswell performance with 6/8 core chips for cheaper money could be viable in the DX12/Vulcan world. Fewer cores for the same performance is always better. Games that don't thread well will always run like garbage, and even games that do thread highly will perform better on fewer/stronger cores due to stuff like less lock contention. DX12/Vulkan is going to force most people to use a big-budget engine that has better-skilled developers, so I guess that means we'll probably see games that thread better in the DX12/Vulkan future, but there's only so far you can go on that. Even highly threaded games tend to lean on one core real hard - some of that is due to the difficulties of threading with current-gen graphics APIs, but there are just some inherent bottlenecks to the game loop. You can pick up a i5-4690K and a MSI Z97 PC Mate for $265 out the door at Microcenter. To be appealing given power usage and the fact that each core is weaker it'd probably have to be down in the $175 range or lower. Factor in $75 for the motherboard manufacturer and you don't leave much room for AMD. Let alone if still you need to bundle it with a water cooler. Paul MaudDib fucked around with this message at 00:53 on May 7, 2015 |
|
#
?
May 7, 2015 00:47
|
|
|
Paul MaudDib posted:Fewer cores for the same performance is always better. Games that don't thread well will always run like garbage, and even games that do thread highly will perform better on fewer/stronger cores due to stuff like less lock contention. I imagine a generation of consoles that have a large number of weak x86 cores is going to continue pushing development there, isn't it?
|
|
#
?
May 7, 2015 00:50
|
|
|
Paul MaudDib posted:Fewer cores for the same performance is always better. Games that don't thread well will always run like garbage, and even games that do thread highly will perform better on fewer/stronger cores due to stuff like less lock contention. We'll see when DX12/Vulcan games start hitting the market. The early synthetic benchmarks indicate a significant boost going from 4 to 6/8. However, they're only synthetic and we won't know until the real thing is running around.
|
|
#
?
May 7, 2015 00:51
|
|
|
El Scotch posted:We'll see when DX12/Vulcan games start hitting the market. The early synthetic benchmarks indicate a significant boost going from 4 to 6/8. However, they're only synthetic and we won't know until the real thing is running around. Synthetic benchmarks also show perfect scaling from 4-way Crossfire/SLI, which doesn't show up in the real world either. Having a game that responds to player interaction (and maybe even talks to stuff on the internet) is way different than running FireStrike. You can pick up a i5-4690K and a MSI Z97 PC Mate for $265 out the door at Microcenter. To be appealing given power usage and the fact that each core is weaker it'd probably have to be down in the $175 range or lower. Factor in $75 for the motherboard manufacturer and you don't leave much room for AMD. Let alone if still you need to bundle it with a water cooler. If DX12 games scale great it might be worth it at $200 or so.
|
|
#
?
May 7, 2015 00:57
|
|
|
Killer robot posted:I imagine a generation of consoles that have a large number of weak x86 cores is going to continue pushing development there, isn't it? I'd think that if AMD was going to get a boost from the fact that their processors are in both current-gen consoles, it would already be showing. Your average console runs on a 1.6 GHz APU and it hasn't made PC games perform vastly better. I think the way this usually tends to play out is that console devs take a PC gaming engine and start stripping features and detail and tuning to the hardware until they can get it to run, rather than engineering something better from the ground up. That poo poo takes too long and if the game isn't in stores by Christmas then your publisher is going to cut you off. It would be sweet to spend a couple years writing an engine that could push 144 FPS on a 16-core potato but the priority tends to be writing games. Paul MaudDib fucked around with this message at 01:15 on May 7, 2015 |
|
#
?
May 7, 2015 01:10
|
|
|
Paul MaudDib posted:Synthetic benchmarks also show perfect scaling from 4-way Crossfire/SLI, which doesn't show up in the real world either. Having a game that responds to player interaction (and maybe even talks to stuff on the internet) is way different than running FireStrike. Zen has a max TDP of 95W, how would this require a water cooler? Fairly sure AMD can hit the competitive price point if they can get a performing product.
|
|
#
?
May 7, 2015 01:12
|
|
|
FaustianQ posted:Zen has a max TDP of 95W, how would this require a water cooler? Fairly sure AMD can hit the competitive price point if they can get a performing product. There's no official mention of what TDP Zen is aiming for.
|
|
#
?
May 7, 2015 01:31
|
|
|
El Scotch posted:We'll see when DX12/Vulcan games start hitting the market. The early synthetic benchmarks indicate a significant boost going from 4 to 6/8. However, they're only synthetic and we won't know until the real thing is running around. You'll still have to deal with games that were released before DX12 and Vulcan were a thing. If ZEN is around Haswell level of performance, that'd be great, but lets not overhype it.
|
|
#
?
May 7, 2015 02:27
|
|
|
GrizzlyCow posted:If ZEN is around Haswell level of performance, that'd be great, but lets not overhype it. Come what come may Time and the hour runs through the roughest day
|
|
#
?
May 7, 2015 06:01
|
|
|
|
| # ? Jan 25, 2025 06:44 |
|
|
Ragingsheep posted:Well they don't have to worry about that problem anymore. SwissArmyDruid posted:Motherfucker, seriously?! Still good!  "You might recall that AMD killed off their SeaMicro systems products a few weeks ago. This was explained by Lisa Su as being a market that wasn’t growing and because AMD wasn’t a systems company, both of which we agree with. The thing lost on most pundits was the product itself was knifed, the underlying tech wasn’t. That will be rolled into the forthcoming server SoC as their fabric, and it will live on, just not as a pre-rolled package." http://semiaccurate.com/2015/05/06/amd-puts-goodies-amid-vague-terms/ Even in the depths of my despair, my precognitive abilities have not failed me! (I am not actually precognitive) SwissArmyDruid fucked around with this message at 10:57 on May 7, 2015 |
|
#
?
May 7, 2015 10:54
|
|