|
AFAIK the APU line is going to be called Athlon, wendell dropped a hint months ago.
|
# ? Jul 14, 2017 20:04 |
|
|
# ? Apr 27, 2024 01:53 |
|
Of course it is, because if there's one thing that AMD does that Intel doesn't, it's a coherent product stack naming scheme.
|
# ? Jul 14, 2017 20:12 |
|
Aww man I was hoping that Athlon would be a name for a higher end part. I want to feel like I'm back to the days of building a super fancy new Athlon based gaming PC.
|
# ? Jul 14, 2017 20:14 |
|
It feel like "Pentium" at this point, even though actual Ryzen-based APUs are probably going to reach up into the lower end of the R5 realm at the top end, (maybe?) as in "a name that is so intrinsically linked to the identity of the company, that its continual reuse is mandated."
|
# ? Jul 14, 2017 20:18 |
|
Nothing's stopping you from just ebaying for some old Athlon T-bird or XP case stickers and doing a retro build calling your r7 1700 an Athlon XP 1700+....
|
# ? Jul 14, 2017 20:28 |
|
NewFatMike posted:R3 hits July 27th, but still no APU information. I had thought for sure R3 was going to have integrated graphics, but it looks like they're just 4C/4T SKUs. The leaked Raven Ridge was a single CCX (4C/8T) with a 704SP graphics unit on-die. Just like Bristol Ridge, this will get cut down/clocked down to hit the mobile power power targets, but will all be the same die. When it actually shows up is another question. If its competitive in mobile, all of the dies will go there for a while cause it has tastier profit margins and a much bigger market.
|
# ? Jul 14, 2017 20:51 |
|
SamDabbers posted:Ryzen owns for multitasking Checking in two months later; Ryzen is still cool and good and has only gotten better with each BIOS and microcode update. Verdict: extremely satisfied
|
# ? Jul 14, 2017 20:59 |
|
SamDabbers posted:Checking in two months later; Ryzen is still cool and good and has only gotten better with each BIOS and microcode update. Feels good.
|
# ? Jul 14, 2017 21:13 |
|
Measly Twerp posted:Feels good. What are you running?
|
# ? Jul 14, 2017 21:14 |
|
Subjunctive posted:What are you running? Compiling Linux for dramatic effect. This is more normal:
|
# ? Jul 14, 2017 21:19 |
|
Yeah, OK.
|
# ? Jul 14, 2017 21:24 |
|
Funny thing, over here in Croatia, it's near impossible to score 3200MHz RAM because as soon as Ryzen appeared, buyers snapped up all the fastest RAM sticks and now nobody has them in stock - or those retailers that do have them in stock, either offer non-Ryzen certified modules or demand seriously exorbitant prices for certified sticks. I'm not surprised though, nobody is going to buy a new R7 system just to put slow 2400MHz memory in it. I hope the situation with RAM will be better by late fall when I intend to upgrade.
A SWEATY FATBEARD fucked around with this message at 08:17 on Jul 15, 2017 |
# ? Jul 15, 2017 08:15 |
|
Looks like somebody leaked a entry level TR 1900X on reddit. The pic was deleted but showed 8C/16T across two dies, 180W TDP, quad-channel, 64 PCIe lanes and likely double the L3 cache (32MB). MSRP $50 higher than the 1800X. It'll be interesting to see how the two dies connected by CCX compare to a 1800X.
|
# ? Jul 15, 2017 09:16 |
|
What could potentially be expected out of a "Zen+" revision of Ryzen chips besides slightly better clock speeds and maybe an iGPU or some similar feature? Is the revision potentially worth waiting half a year for or does Ryzen 7 have all bases covered relatively well, so to speak? Ryzen PRO looks to be identical to Ryzen, but with a separate co-processor for some security features a normal person wouldn't bother to go near. I have no idea what I'm trying to ask here, but I'm basically skirting around the "Does Ryzen's dependency on fast memory clocks possibly diminish (significantly) with a hardware refresh?" question. Looking to upgrade my PC and while I had decided to put it off until next year to see what cool new chips will be announced after Skylake X, Ryzen 3 and Threadripper come out, 7740X's and 7900X's benchmark numbers being on the same ballpark of AMD's 1800X, 1700X and even the value-minded 1600X chips are making it hard to dodge the idea that I will not be seeing a massive appreciable jump in performance for years to come, if ever, so waiting for that specifically is starting to feel pretty pointless. My sole concern on adopting AMD at this point, is that need for 3000MHz+ memory in order to not end up with a bottlenecked system by 10-25% margins, though from reading the last pages of the thread, this is apparently getting better and better through BIOS updates? I've always been a cheapo-RAM sorta guy and having to go for the gamer stuff with the heat spreaders that look like you could slice someone's throat with is a whole new scary world for me. OSad fucked around with this message at 11:02 on Jul 15, 2017 |
# ? Jul 15, 2017 10:41 |
|
The Zen design has a lot of future improvements they could do, FMA and AVX could improve, running the fabric asynchronously to IMC or at better dividers. overall integer performance, moving to 6 core CCX, etc. If you're just talking about a the current designs maturity, looking at Polaris and Vega indicates something like Pinnacle Ridge might have slightly higher clock speeds, at most 200-300Mhz, and it's unlikely the IF being tied to the IMC will change in a "refresh". Ryzen itself doesn't have any gamebreaking bugs in hardware, everything people are having issues with so far seems to be a software issue, or firmware one. If you want to wait and see if the refresh brings what you want I'd suggest it, as it'll give more time for motherboards to mature and diversify. That and DDR4 prices are crazy.
|
# ? Jul 15, 2017 11:15 |
|
The main reason why Zen is so memory speed dependent is that the infinity fabric (the bus that connects the cores) scales off RAM speed. Zen+ may introduce its own infinity fabric multiplier so the interconnect runs faster with slow RAM. Intel's ringbus already does that and the knowledgeable folks at anandtech agree that there's no good reason why Zen doesn't, short of "AMD ran out of time for the Ryzen launch". The nice thing about AM4 in is current state is that Zen+ is (almost) guaranteed to run on current motherboards so you wouldn't lose much by building a system now and upgrading later. Ryzen as a platform appears to be surprisingly reliable and solid considering they started from scratch. I build a R7 1700 system myself and the issues I ran into were either component manufacturer related defects or software bugs. If you're interested in gaming then it may be worth waiting for the 6C/12T Coffee Lake launch; if you're doing a lot of heavy duty content creation then the entry level Threadripper would give you an upgrade path to at least 16C/32T, potentially even 32C/64T in the future. For everything else it is hard to go wrong with any of the 1600/1700 X and non-X CPUs. e: typo eames fucked around with this message at 11:59 on Jul 15, 2017 |
# ? Jul 15, 2017 11:30 |
|
Actually yea that's a good point, TR4 will likely have similar longevity as AM4, compared to the near yearly change of sockets in Intel HEDT.
|
# ? Jul 15, 2017 11:53 |
|
On the other hand, depending on what you want to have running on your board, you might want to wait. That said, if the supposed 8C/16T TR clocks high enough, I might get tricked into upgrading the CPU eventually. Now a cheap TR waiting for a TR+ or TR2.eames posted:Zen+ may introduce its own infinity fabric multiplier so the interconnect runs faster with slow RAM. Intel's ringbus already does that and the knowledgeable folks at anandtech agree that there's no good reason why Zen doesn't, short of "AMD ran out of time for the Ryzen launch". Combat Pretzel fucked around with this message at 12:42 on Jul 15, 2017 |
# ? Jul 15, 2017 12:34 |
|
How did Intel respond to accusations of just gluing a CPU into a socket?
|
# ? Jul 15, 2017 12:46 |
|
FaustianQ posted:The Zen design has a lot of future improvements they could do, FMA and AVX could improve, running the fabric asynchronously to IMC or at better dividers. overall integer performance, moving to 6 core CCX, etc. If you're just talking about a the current designs maturity, looking at Polaris and Vega indicates something like Pinnacle Ridge might have slightly higher clock speeds, at most 200-300Mhz, and it's unlikely the IF being tied to the IMC will change in a "refresh". eames posted:Zen+ may introduce its own infinity fabric multiplier so the interconnect runs faster with slow RAM. Intel's ringbus already does that and the knowledgeable folks at anandtech agree that there's no good reason why Zen doesn't, short of "AMD ran out of time for the Ryzen launch". Well, I'll just wait it out then. Just the idea of not having to pay too much for RAM is enough to hold my hand on this one, I certainly want at least 32 gigs of the stuff as I do occasional content creation and I've hit the paging file a few too many times to know that accessing virtual memory from the disk drains your sanity rather fast. Combat Pretzel posted:On the other hand, depending on what you want to have running on your board, you might want to wait. That said, if the supposed 8C/16T TR clocks high enough, I might get tricked into upgrading the CPU eventually. Now a cheap TR waiting for a TR+ or TR2. This is also an interesting idea. Theorizing that the larger chip size could lead to better heat dissipation and better overclocking, an 8C TR could OC like a champ and bring stuff like 60 PCI-E lanes and lower TDP to the table (not that I would ever need to use 60 whole PCIe lanes). Maybe I'm just letting myself get taken on the hype train a bit, but the $800 12C/24T Threadripper model sounds just affordable enough that I would seriously consider saving up a bit more cash to get it over the 1800X, just to have a platform I can potentially upgrade to the super-beefier core-count models later. I'm very excited to look at TR benchmarks when they come out, they're probably not going to set the world on fire, but based on some early numbers AMD is putting out, TR looks like another killer value proposition which I'm all for.
|
# ? Jul 15, 2017 13:06 |
|
DDR4 RAM is super expensive in general right now and it's just been so hard for manufacturers to keep up with demand. Hopefully relief will come sooner than later, but the list of smart devices that need memory doesn't seem to be going down right now.
|
# ? Jul 15, 2017 13:20 |
|
Sidesaddle Cavalry posted:DDR4 RAM is super expensive in general right now and it's just been so hard for manufacturers to keep up with demand. Hopefully relief will come sooner than later, but the list of smart devices that need memory doesn't seem to be going down right now. Is the DDR4 expense just a matter of demand outstripping supply then? Are some manufacturers off line because they're switching over from DDR3?
|
# ? Jul 15, 2017 19:32 |
|
WAR DOGS OF SOCHI posted:Is the DDR4 expense just a matter of demand outstripping supply then? Are some manufacturers off line because they're switching over from DDR3? Yes and kindof. Apple is sucking up supply for the iPhone 8 launch and micron had an "incident" at one of their fabs that took it offline for a few weeks or something. Among other reasons...
|
# ? Jul 15, 2017 20:10 |
|
DRAM/Flash is also completely price fixed (still), so theres that.
|
# ? Jul 15, 2017 20:29 |
|
Making nand is printing money at the moment since the demand is insane and supply is constrained to 3-4 makers Toshiba's entire net worth is their nand division
|
# ? Jul 15, 2017 20:31 |
|
Cygni posted:DRAM/Flash is also completely price fixed (still), so theres that.
|
# ? Jul 15, 2017 22:17 |
|
DDR4 was cheap last year because Skylake was so ho-hum and also had DDR3 support, while the suppliers overproduced RAM hoping Win 10 would spike demand which never happened thanks to the free upgrade. It also doesn't help consumer desktop parts are now at the bottom of the market totem pole. What you want to hope for are the PRC semicon foundries coming online soon and opening the supply floodgates, like they did with li-ion batteries to the point where Samsung, LG, Sony etc are now abandoning the consumer battery market.
|
# ? Jul 16, 2017 03:48 |
|
eames posted:Looks like somebody leaked a entry level TR 1900X on reddit. If such a product exists wouldnt the extra 32mb be more like a l4 then extra l3 given it will be on the other 'die'?
|
# ? Jul 16, 2017 04:48 |
|
Scarecow posted:If such a product exists wouldnt the extra 32mb be more like a l4 then extra l3 given it will be on the other 'die'?
|
# ? Jul 16, 2017 04:53 |
|
Anime Schoolgirl posted:1500x has access to the full 16mb cache despite only having 4 active cores so probably Yeah but ripper is 4/4+4/4 so would a 8 core ripper be 4/0+4/0 or 4/4+0/0? I kinda want to think the former would be the case so that would give the 4/0 set up 16mb each so it could possibly perform better then a 1800x?
|
# ? Jul 16, 2017 05:00 |
|
A theoretical 8 core TR would be 2/2+2/2. You could also do a comical 4 core with 1/1+1/1, which would join the Epyc 7251 8 core (1/1+1/1+1/1+1/1) in "this is for a very specific use case" territory, ha. Lots of cache/memory bandwidth, lots of PCIe, not a lot of raw proc power.
|
# ? Jul 16, 2017 05:11 |
|
Yeah, I think that would be the "just enough to light up the platform" chip for people who need as many GB of memory and/or PCIe lanes as they can get at a given price point and don't care about the actual CPU performance much e.g. Xeon E5-2603.
|
# ? Jul 16, 2017 05:14 |
|
Scarecow posted:Yeah but ripper is 4/4+4/4 so would a 8 core ripper be 4/0+4/0 or 4/4+0/0? I kinda want to think the former would be the case so that would give the 4/0 set up 16mb each so it could possibly perform better then a 1800x? TR is two dies, so it would be 2+2 on each die. e:fb
|
# ? Jul 16, 2017 08:52 |
|
SwissArmyDruid posted:TR is two dies, so it would be 2+2 on each die. but isnt it 2 CCX per main die? so its 2CCX or 8 cores in a 4/4 split per CCX?
|
# ? Jul 16, 2017 09:02 |
|
A die is a CCX is 4 cores plus 4 cores. A fully-enabled TR part is two dies is two CCSes, which is 4 cores + 4 cores on one die, and 4 cores + 4 cores on the other, giving 16c/32t. A theoretical 8c/16t would have each of those numbers divided in half. 2 cores + 2 cores on one die, and 2 cores + 2 cores on the other. SwissArmyDruid fucked around with this message at 09:52 on Jul 16, 2017 |
# ? Jul 16, 2017 09:36 |
|
SwissArmyDruid posted:A die is a CCX is 4 cores plus 4 cores. A CCX is only four physical cores. Ryzen is made up of two CCXs, for an eight core maximum. Thread Ripper will be made up of four CCX units. A singular die is two CCXs.
|
# ? Jul 17, 2017 06:08 |
|
Does AM4 run 2x16GB DDR4-3200 reliably now?
|
# ? Jul 17, 2017 12:13 |
Palladium posted:Does AM4 run 2x16GB DDR4-3200 reliably now? That's too vague to answer, if your asking if you can grab any random combo of mobo and DDR4-3200 RAM and have everything work fine then no, you still need to make sure your mobo and RAM will work together at the listed speeds, also probably do some BIOS updates. If you do your homework you should be able to get a combo like that working but people still run into problems occasionally.
|
|
# ? Jul 17, 2017 12:42 |
|
SlayVus posted:A CCX is only four physical cores. Ryzen is made up of two CCXs, for an eight core maximum. Thread Ripper will be made up of four CCX units. A singular die is two CCXs. Why hello there, Mr. Shoe. You look mighty delicious right now, why don't we give you a taste?
|
# ? Jul 17, 2017 13:28 |
|
|
# ? Apr 27, 2024 01:53 |
|
Palladium posted:Does AM4 run 2x16GB DDR4-3200 reliably now? Which board? Figure that out and search the web cache for cpuz validations. This isn't ideal, but it's a start https://www.google.ca/?q=+3200+ryzen+site:http://valid.x86.fr/
|
# ? Jul 17, 2017 14:29 |