|
That thing is going to be loud. Why do they always mess up the cooling aspect of SFF PCs.
|
#
?
Jan 8, 2018 11:41
|
|
|
|
| # ? Apr 25, 2024 07:45 |
|
|
People with AMD processors should keep an eye out for UEFI updates. Just as with the Intel Managemen Engine, AMD's security features appears to be not entirely secure.
|
|
#
?
Jan 8, 2018 12:50
|
|
|
Drakhoran posted:People with AMD processors should keep an eye out for UEFI updates. Just as with the Intel Managemen Engine, AMD's security features appears to be not entirely secure. people with AMD processors can disable PSP/TPM in the bios
|
|
#
?
Jan 8, 2018 13:04
|
|
|
Palladium posted:2200G/2400G, aka "who is gonna give a poo poo to the 1300X/1500X anymore". I mean, they probably *are*. Now that we seem to have what the APU was supposed to do half a decade ago with Bulldozer, APUs exist to do three things: *Claw market share away from Intel *Claw market share away from Nvidia. *Do the above at the same time. Since we've seen that Vega does... for now, let's say it does "okay" at the low end (at least until we get more SKUs with APUs in them for benchmarking to get a better picture), any APUs they sell are definitively one less Intel sale, and a good chance it's one fewer Nvidia sale of like, a 1050Ti or some poo poo. SwissArmyDruid fucked around with this message at 13:13 on Jan 8, 2018 |
|
#
?
Jan 8, 2018 13:10
|
|
|
Klyith posted:people with AMD processors can disable PSP/TPM in the bios Not if their motherboard manufacturer doesn’t support it.....MSI.
|
|
#
?
Jan 8, 2018 13:25
|
|
|
How's asrock AMD bioses these days? Stoked for April.
|
|
#
?
Jan 8, 2018 16:06
|
|
|
Ugh, those slides imply Zen 2 TR to be H2/Q4 2019 >:[
|
|
#
?
Jan 8, 2018 16:34
|
|
|
SwissArmyDruid posted:I mean, they probably *are*. Might as well sell the $100 APU, when nobody really gives a poo poo about the RX460/560. Shame about the DDR4 prices though.
|
|
#
?
Jan 8, 2018 17:00
|
|
|
The 2200 and 2400G seem to be perfect for Cybercafes, based on Bristol Ridge performance they should fantastic for "epsorts" titles at medium 720p/1080p. I mean the RX560 can do high/epic and Raven Ridge will always chug on poo poo like PUBG but the solution then is to just add in a 1060/1070. Because of this it kind of makes it questionable why AMD would continue selling 4C/8T non APU parts, what's the target market besides extreme budget builds? On another note, we know that AMD will be doing their own version of the 8809G, the question I have is whether or not anyone thinks it'd be feasible for AMD to make it socketed for TR4, and what do you think the lower bound would be on a TR4 motherboard?
|
|
#
?
Jan 8, 2018 17:44
|
|
|
GRINDCORE MEGGIDO posted:How's asrock AMD bioses these days? Solid AF. Have their Fatality x370 pro gaming board. I was impressed when I could just flash the bios from the Internet within the BIOS itself. 
|
|
#
?
Jan 8, 2018 17:54
|
|
|
Llamadeus posted:Just announced! Naturally, it's Asrock: https://www.asrock.com/news/index.asp?iD=3888 Before I clicked on that I figured the socket would take up about 1/3 of the board and I was not too wrong SlayVus posted:4 DIMM slots, 8 SATA slots, 3 m.2 slots and 1 u.2, 3 front USB headers, 3 PCI-e 16x slots, dual Intel Gigabit, dual band WiFi, at least 8 power phases. Power phases go all the way up to 11 https://techreport.com/news/33040/asrock-x399m-taichi-squishes-threadripper-into-microatx
|
|
#
?
Jan 8, 2018 18:00
|
|
|
GRINDCORE MEGGIDO posted:How's asrock AMD bioses these days? Good if a little slow. Still waiting for a release to fix the X370 Tachi’s 1.55 GHZ bug. There’s a beta floating around that came from some German site that apparently fixes it but I’ll wait until its officially released.
|
|
#
?
Jan 8, 2018 19:29
|
|
|
RAVEN RIDGE posted:They're also unlocked so are overclockable and demonstrations at a CES Tech Day event showed overclocked CPU speeds up to 4.2GHz and Vega GPU overclocking yielding close to 50% performance boosts in games too. quote:Hardware Unboxed guy said AMD demonstrated it [GPU from APU] clocking up to 1750MHz. But, interestingly- very low stock clocks, similar to U-series- 1250MHz and 1100MHz 
|
|
#
?
Jan 8, 2018 20:26
|
|
|
Hi thread  What do you all think about the Ryzen 1700 at $250 + a A350 Asrock mobo at $88? Should I wait for AMD's April refresh, or splurge now? This will be a turnkey, as I have everything else needed for this build.
|
|
#
?
Jan 8, 2018 20:39
|
|
|
A Bad King posted:Hi thread April is not that far, and if you haven't invested now, better to wait for the 400 series boards and 2000 series processors. Unless you need it now, IMHO better to wait.
|
|
#
?
Jan 8, 2018 20:50
|
|
|
GamersNexus video: https://www.youtube.com/watch?v=ea5CDTUivys Raven Ridge APU will have the Vega cores linked via Infinity Fabric. It will be interesting to compare performance with the 8809G, where that connect is via PCIe 8x.
|
|
#
?
Jan 8, 2018 22:36
|
|
|
FaustianQ posted:April is not that far, and if you haven't invested now, better to wait for the 400 series boards and 2000 series processors. Unless you need it now, IMHO better to wait. $250 is a drat good price though -- the question is, why such a deep discount and what does it mean about how good zen+ will be? one possibility is that zen+ is gonna be a great improvement (like into the 4.5ghz zone), and the cut now is reflecting the new price/performance to expect from that. they're selling old zen for a song because they're about to get totally outclassed. the other possibility is that zen+ is a more modest bump and the discount is more to clear inventory because they overproduced in late 2017. also, they were handing out 8-core 1600s recently, the cheap 1700 may just be trying to push people into just buying a drat 8-core. they seem to have plenty! so in the first case you get a cheap 8 core now but miss out on the future 6 core with much better single-thread performance. so for a gamer who cares about the comparison to coffee lake, definitely wait until we start getting news or previews of the new chip. a non-gamer can either buy now or wait a bit, depending on what type of workload you have and how much you need the new pc right away. in the second scenario the 1700 sale is a better deal than where zen+ will be introduced at and buying ahead of zen+ is actually the smart move.
|
|
#
?
Jan 8, 2018 23:40
|
|
|
SwissArmyDruid posted:GamersNexus video: https://www.youtube.com/watch?v=ea5CDTUivys Wondering if by being linked through Infinity Fabric the dGPU can offload rendering work to the iGPU, like say physics or anti-aliasing? Not a huge potential improvement, maybe 20%?
|
|
#
?
Jan 8, 2018 23:59
|
|
|
FaustianQ posted:Wondering if by being linked through Infinity Fabric the dGPU can offload rendering work to the iGPU, like say physics or anti-aliasing? Not a huge potential improvement, maybe 20%? That's already possible with any combination of dGPU and iGPU using DX12 unlinked multi-adapter mode, the limiting factor is that no engine developer gives a poo poo about supporting such an esoteric setup.
|
|
#
?
Jan 9, 2018 00:08
|
|
|
What, realistically, is expected of the respin, is the new process likely to really allow clocks of around 4.3ish, something like that? Wonder if the memory controller is happier with faster RAM speeds or not.
|
|
#
?
Jan 9, 2018 00:08
|
|
|
GRINDCORE MEGGIDO posted:What, realistically, is expected of the respin, is the new process likely to really allow clocks of around 4.3ish, something like that? I think a reasonably optimistic guess would put most chips at 4.3-4.4 with a bit of extra voltage and golden chips (threadripper-level bins) are 4.5. This is purely pulled out of my rear end however, and it depends on how much you take GloFo's "15% performance improvement" at face value, how much voltage you are willing to apply, and what clocks you define as the starting point. It would be unrealistic to expect more than 4.5 average and 4.7 as a golden chip/threadripper, that's the absolute best case. I'm guessing the memory controller might have better stability, yeah. I bet the segfault bug gets fixed too. Paul MaudDib fucked around with this message at 00:25 on Jan 9, 2018 |
|
#
?
Jan 9, 2018 00:16
|
|
|
A 4.7ghz TR makes my nipples rock hard god drat i want it
|
|
#
?
Jan 9, 2018 00:28
|
|
|
Scarecow posted:A 4.7ghz TR makes my nipples rock hard god drat i want it Yeah my dick's getting real hard for it too, ripping multiple threads now, come on put that pump on my hot smooth body
|
|
#
?
Jan 9, 2018 00:40
|
|
|
repiv posted:That's already possible with any combination of dGPU and iGPU using DX12 unlinked multi-adapter mode, the limiting factor is that no engine developer gives a poo poo about supporting such an esoteric setup. Of course, but what I am talking about is whether that can be done at a hardware level. Like what makes IF specifically better than using a PCIE connection considering it's slower. Maybe an advantage is that the dGPU can "cannibalize" the iGPU for latency insensitive workloads? Maybe the iGPU gets access to the HBM2 as well, in which case huge performance uplift even when running low power? Also it looks like both Intel and AMD solutions have the same Z height, so looks like AMD has access to EMIB ("You get Raja, we get EMIB, joint project for two years?"). If so, does that mean AMD can reduce latency in EPYC/Rome between dies?
|
|
#
?
Jan 9, 2018 00:55
|
|
|
FaustianQ posted:Of course, but what I am talking about is whether that can be done at a hardware level. Like what makes IF specifically better than using a PCIE connection considering it's slower. Maybe an advantage is that the dGPU can "cannibalize" the iGPU for latency insensitive workloads? Maybe the iGPU gets access to the HBM2 as well, in which case huge performance uplift even when running low power? How would the hardware know what constitutes a latency insensitive workload? It doesn't have that kind of high level context. Without engine developers guiding the iGPU offload the only alternative I see is driver fuckery with game specific profiles, which is exactly what AMD has been trying to move away from.
|
|
#
?
Jan 9, 2018 01:10
|
|
|
GRINDCORE MEGGIDO posted:What, realistically, is expected of the respin, is the new process likely to really allow clocks of around 4.3ish, something like that? IMHO I really doubt it, it's a process tweak and a design tweak. If there was that type of potential available, AMD & Gloflo really hosed it the first time around. Which is why I think the most likely reason the 1700 is selling for $250 is that their yield is so good they are oversupplied on functional 8 core chips. There's obvious evidence to support that. And that could totally reset when they switch to zen+. Meaning that buying now isn't totally crazy if the $250 in april is back to buying a 6 core.
|
|
#
?
Jan 9, 2018 01:13
|
|
|
That's a Microcenter price. It isn't indicative of normal reality. I want to try and wait for April but I don't know when I'll be able to buy 8c/16t again for under $300.
|
|
#
?
Jan 9, 2018 01:25
|
|
|
I'm glad I saw this thread and that Zen+ is coming before I dropped money on an 1800x... or rather the ram for the fuckin' thing. Thanks guys.
|
|
#
?
Jan 9, 2018 01:33
|
|
|
I think its worth noting that when they say "infinity fabric" connects the GPU and CPU, they are talking about the internal on-die interconnect. It's a 1 die package, unless something weird has changed. "Infinity Fabric" is the lovely marketing name they are using for like 5 different interlink speeds, essentially the HT protocol over a PCIe lane of varying widths (in some configurations). "Infinity Fabric" can be insanely fast and wide, tight and fairly slow, on die, off die on package, or off package completely. Cygni fucked around with this message at 01:44 on Jan 9, 2018 |
|
#
?
Jan 9, 2018 01:42
|
|
|
Ranzear posted:I'm glad I saw this thread and that Zen+ is coming before I dropped money on an 1800x... or rather the ram for the fuckin' thing. Thanks guys. There’s no guarantee the RAM prices will drop by the time Zen+ comes out unless China stomps down hard and quick.
|
|
#
?
Jan 9, 2018 01:57
|
|
|
The ram I picked was $100 more than the sticker on the 1800X is all I was getting at. Sure I could reuse it on a Zen+ but I'm having Phenom flashbacks here.
|
|
#
?
Jan 9, 2018 02:01
|
|
|
GRINDCORE MEGGIDO posted:What, realistically, is expected of the respin, is the new process likely to really allow clocks of around 4.3ish, something like that? Around there. The real gain, at least for gaming, will be if they can get the single core boost up higher. One of the reasons why Coffee Lake does so well on gaming is that it will boost one core up +1GHz. If the 2700/2800 can get a single core to push 5GHz they're neck and neck for single threaded performance.
|
|
#
?
Jan 9, 2018 02:02
|
|
|
Xae posted:The real gain, at least for gaming, will be if they can get the single core boost up higher. One of the reasons why Coffee Lake does so well on gaming is that it will boost one core up +1GHz. And as a bonus, it doesn't take a whole lot of extra juice at all to get all core +1 GHz.
|
|
#
?
Jan 9, 2018 02:22
|
|
|
An 4.7GHz TR would be hilarious, but there's still the thing with the IF. Is there 3200MHz ECC RAM yet? I thought Crucial said they'd do it.
|
|
#
?
Jan 9, 2018 05:52
|
|
|
Combat Pretzel posted:An 4.7GHz TR would be hilarious, but there's still the thing with the IF. Is there 3200MHz ECC RAM yet? I thought Crucial said they'd do it.
|
|
#
?
Jan 9, 2018 05:53
|
|
|
Anime Schoolgirl posted:samsung is already making 3200mhz dies but all of them are going straight into HPC and are $$$$$$custom orders I have a TR 1950x with 64GB (4x16GB) of Samsung B-die ECC RAM, and it works fine. Mem overclocked just fine, bumped it from 2400 to 2933 or so with a slight voltage bump. System sees that it's ECC and if I go full retard on the OC, you can see the system log start to rapidly poo poo itself with single bit ECC errors, so it works fine. Asrock X399 taichi, 1950x, the Liqtech TR4 360mm, and my usual pile of add-in cards. I'm not looking forward to a die shrink/refresh for my TR, but holy poo poo, getting an extra 400-700 Mhz and possibly 50% more cores would be amazing.
|
|
#
?
Jan 9, 2018 06:16
|
|
|
More fat for the Intel fire. https://twitter.com/qrs/status/950462488348446721
|
|
#
?
Jan 9, 2018 06:23
|
|
|
More fat for the AMD fire. https://www.theverge.com/2018/1/9/16867068/microsoft-meltdown-spectre-security-updates-amd-pcs-issues quote:“Microsoft has reports of customers with some AMD devices getting into an unbootable state after installing recent Windows operating system security updates,” says a Microsoft spokesperson. “After investigating, Microsoft has determined that some AMD chipsets do not conform to the documentation previously provided to Microsoft to develop the Windows operating system mitigations to protect against the chipset vulnerabilities known as Spectre and Meltdown.”
|
|
#
?
Jan 9, 2018 13:23
|
|
|
Rastor posted:More fat for the AMD fire. The processors in question were Athlons and Semprons. Not sure why the Verge felt the need to omit that...
|
|
#
?
Jan 9, 2018 13:29
|
|
|
|
| # ? Apr 25, 2024 07:45 |
|
|
Because it's the verge.
|
|
#
?
Jan 9, 2018 13:50
|
|
