|
kirtar posted:There was that AMD presentation a while back that suggested vSOC 1.1V and the RAM voltage 1.4-1.5V. I'm running dual rank B-die (G.Skill F4-3200C14D-32GTZKY) at the same timings and frequency without any voltage increase, so I'm surprised that there is a problem on single rank B-die. Video below in case you feel like watching. Chiming in to add that I did set the RAM voltage to 1.4, which is what the sticker on the box told me it wanted and it seems fine though I've only tested with memtest so far because I haven't quite pulled the trigger on the storage and don't feel like installing anything temporarily. We'll see soon if that falls apart under 'real' load, I guess 
|
#
?
Sep 1, 2017 14:07
|
|
|
|
| # ? Apr 25, 2024 18:10 |
|
|
This is a good a time as any to ask the embarrassing question: what is 'pass thru' wrt virtualization?
|
|
#
?
Sep 1, 2017 15:45
|
|
|
Instead of virtualizing a resource through an abstraction layer, you dedicate its resources and hardware addresses exclusively to a single VM. In most cases, you have a second GPU and you dedicate that directly to a single VM so it doesn't have to use a virtualized GPU driver. This started out back in the day with being able to passthrough something like a dedicated NIC or SCSI card to a single VM on your hypervisor. I did it for a while to virtualize an old backup server with a direct attach SAS tape library.
|
|
#
?
Sep 1, 2017 15:49
|
|
|
Ah, got it. Thank you!
|
|
#
?
Sep 1, 2017 16:29
|
|
|
Mindfactory (very large german retailer) displays their sale stats on the product pages and people figured out that AMD overtook Intel this month in both CPUs sold and CPU revenue  http://imgur.com/a/O2O5C
|
|
#
?
Sep 1, 2017 19:51
|
|
|
Wow 7600K got wrecked.
|
|
#
?
Sep 2, 2017 17:23
|
|
|
I really, really want to replace my 3770k with a 1700 or 1700x, but I can't stand how expensive 3200mhz DDR4 is, or how picky Ryzen is about running RAM at rated speeds. For those deeper in this than me, the RAM-not-running-at-rated-speeds issue has been slowly getting better over time, right? I know that each motherboard manufacturer publishes compatibility lists, but sorting through those lists and trying to find the best, cheapest RAM to buy is time-consuming.
|
|
#
?
Sep 2, 2017 17:28
|
|
|
Ryzen's memory controller ages like fine wine
|
|
#
?
Sep 2, 2017 17:38
|
|
|
I've been wanting to get a 1700 also and overclock it, but it's hard with Coffee Lake so close and wanting to see how the two compare...
|
|
#
?
Sep 2, 2017 17:47
|
|
|
SourKraut posted:I've been wanting to get a 1700 also and overclock it, but it's hard with Coffee Lake so close and wanting to see how the two compare... I'm waiting to see what Coffeelake does to Ryzen prices. GRINDCORE MEGGIDO fucked around with this message at 18:39 on Sep 2, 2017 |
|
#
?
Sep 2, 2017 18:28
|
|
|
GRINDCORE MEGGIDO posted:I'm waiting to see what Coffeelake does to Ryzen prices. Yeah, that's part of it too. I don't know how much more AMD could cut prices though? The 1700 already goes down to $270 - would we see it drop even lower, if the i5-8600K comes in at the rumored $250-$280 range? Or would AMD maybe just lower it permanently to $270 or whatever and say "16 threads to 6, single threaded performance good enough!"?
|
|
#
?
Sep 2, 2017 18:44
|
|
|
GRINDCORE MEGGIDO posted:I'm waiting to see what Coffeelake does to Ryzen prices.
|
|
#
?
Sep 2, 2017 19:35
|
|
|
A SWEATY FATBEARD posted:I'm a proud new owner of a 1700X on an ASUS Prime X370-PRO board. The processor is modestly overclocked to 3700 MHz and under full load the temperature doesn't go over 65C. The downside is that the machine is ridiculously picky about RAM, I got me a pair of 3200MHz sticks which max out at 2133MHz, and then I swapped them out for a pair of Ryzen-certified 3200MHz sticks which top out at 2400 MHz instead. I'm not too happy about that, updated the BIOS to no avail. youtube stuttering is less of an indictment of the fx cpu and more an indictment of goog's boneheaded insistence on encoding everything in vp9 when a majority of computers in use don't have a hardware path for it. get h264ify.
|
|
#
?
Sep 2, 2017 23:33
|
|
|
There's some chatter over on the AnandTech forums about primitives binning tuned for Raven Ridge and not Vega. I'm so curious about these APUs. I can't wait to get weird with them.
|
|
#
?
Sep 3, 2017 06:17
|
|
|
Generic Monk posted:youtube stuttering is less of an indictment of the fx cpu and more an indictment of goog's boneheaded insistence on encoding everything in vp9 when a majority of computers in use don't have a hardware path for it. get h264ify. It wasn't just Youtube, attempting to view any HD video while system was busy number crunching in the background lead to stuttering. I blame dire cache mispredictions causing the system to pull data out of system RAM instead from L2 cache. Also, I RMA-d the misbehaving DDR4 memory sticks and the guys at the service center told me that apparently the whole series of these memories are duds - they already had three RMAs of these sticks the past week alone and they agreed to replace them, no questions asked. So it looks like it isn't the processor's fault, G.Skill RAM just aint up to the task it would seem. We'll see how new sticks will perform - and if they too refuse to work as advertised, the guys told me that they'll give me my money back.
|
|
#
?
Sep 3, 2017 12:06
|
|
GAY 4 ORGANS
A SWEATY FATBEARD posted:It wasn't just Youtube, attempting to view any HD video while system was busy number crunching in the background lead to stuttering. I blame dire cache mispredictions causing the system to pull data out of system RAM instead from L2 cache. That RAM has worked for tons of other people, it sounds like they got a bad batch, also the actual memory is made by Samsung, not G.Skill.
|
|
|
#
?
Sep 3, 2017 12:16
|
|
|
I really want to put together a new system with a mini-ITX Ryzen mobo, but right now the reviews on the motherboards that I've been looking at are really really mixed.
|
|
#
?
Sep 3, 2017 14:01
|
|
|
Three-Phase posted:reviews on the motherboards that I've been looking at are really really mixed. That sounds about right for Ryzen, heh.
|
|
#
?
Sep 3, 2017 14:32
|
|
|
Decided to go with a ryzen 7 1700x for my upgrade, now it's time to settle on a motherboard. Do people here have any experience or strong opinions about any of the following: MSI ATX DDR4 AM4 AMD X370 Chipset SATA III ASUS Prime X370-Pro AMD Ryzen AM4 GIGABYTE AORUS GA-AX370-Gaming K7 (rev. 1.0) AM4 AMD ASUS ROG Crosshair VI Hero AM4 AMD X370 SATA
|
|
#
?
Sep 3, 2017 18:37
|
|
|
Wirth1000 posted:That sounds about right for Ryzen, heh. Yeah there was a Gigabyte mobo that would work with R7 but require a firmware update to run R5. It's kinda hard to do a firmware update without a functioning CPU... and I don't think the BIOS is on replaceable DIP socketed chips anymore.
|
|
#
?
Sep 3, 2017 18:41
|
|
|
Three-Phase posted:and I don't think the BIOS is on replaceable DIP socketed chips anymore. That sucks. I was in that situation with phenom 2 and used an eBay bios chip to fix it.
|
|
#
?
Sep 3, 2017 19:21
|
|
|
GRINDCORE MEGGIDO posted:That sucks. I was in that situation with phenom 2 and used an eBay bios chip to fix it. Lots of motherboards still have replaceable bios chips. They are typically 8 pin chips.
|
|
#
?
Sep 3, 2017 20:21
|
|
|
Three-Phase posted:Yeah there was a Gigabyte mobo that would work with R7 but require a firmware update to run R5. The one good thing about my current X299 Asus board (with all its USB issues) is that it has a feature where you can update the BIOS with a USB stick even if there's no CPU in the system, given it needed an update for the Intel Core i7 6800k I bought it with.
|
|
#
?
Sep 4, 2017 02:25
|
|
|
WCCFT article, salt now so you're not salty later: http://wccftech.com/amd-talks-ryzen-zen-2-ipc-clock-speed/quote:Don Woligroski, Desktop CPU Marketing Manager at AMD
|
|
#
?
Sep 5, 2017 08:19
|
|
|
Hhrrrrfgh i can only be this erect for so long
|
|
#
?
Sep 5, 2017 08:24
|
|
|
Clock and IPC bump? Please to be giving soon!
|
|
#
?
Sep 5, 2017 09:01
|
|
|
Even if it was just a clock bump that would shake things up so hard
|
|
#
?
Sep 5, 2017 09:09
|
|
|
AMD isn't that far behind Intel even now, both an IPC and clock bump would put them within margin of error for Coffeelake. If they have anything at all, a leak right now to discourage Coffeelake sales would be a really good marketing move.
|
|
#
?
Sep 5, 2017 09:11
|
|
|
Man, if they get anywhere near the Bulldozer -> Piledriver improvements that happened as their last uArch matured, they'll be in amazing shape.
|
|
#
?
Sep 5, 2017 12:46
|
|
|
We've got the best 2018, don't we folks? I'm excited for Zen 2 or Plus
|
|
#
?
Sep 5, 2017 13:20
|
|
|
Zen2 has always had the potential for greatness. Zen1 is already not half bad, if they can bump the IPC and clocks up a bit, fix the memory controller's touchiness, and get some of the errata fixed then Zen2 could easily be going toe-to-toe with Coffee Lake.
|
|
#
?
Sep 5, 2017 16:14
|
|
|
Running the internal clocks async from the memory controller clock is about the only thing keeping me away. I don't want to sit there and screw around with optimizing the hell out of my memory timings for a bunch of memory bandwidth I can't actually use just to get the inter-core transport running faster.
|
|
#
?
Sep 5, 2017 16:16
|
|
|
Just a reminder that if you go off AnandTech's reporting on the process nodes, Zen 2 won't show up until late 2018/early 2019, which means it will potentially be going against Intel's 10nm Ice Lake and not Coffee Lake. Rumor is that Zen will get a mild refresh in 2018 to go against Coffee Lake, but no process or architectural changes. So maybe a few hundred mhz, but no IPC change. Late '18-early '19 will be when things really get interesting for both companies.
|
|
#
?
Sep 5, 2017 17:06
|
|
|
And judging by the mask count Intel is undergoing one of the biggest rolling shitshows in 10nm and pushing any improvements down the stack to 7nm
|
|
#
?
Sep 5, 2017 17:08
|
|
|
Cygni posted:Just a reminder that if you go off AnandTech's reporting on the process nodes, Zen 2 won't show up until late 2018/early 2019, which means it will potentially be going against Intel's 10nm Ice Lake and not Coffee Lake. Rumor is that Zen will get a mild refresh in 2018 to go against Coffee Lake, but no process or architectural changes. So maybe a few hundred mhz, but no IPC change. Is Zen2 the same as Zen+ and are we still talking about AM4 or is this AM4+? The 2 vs + is making me  here
|
|
#
?
Sep 5, 2017 17:18
|
|
|
Munkeymon posted:Is Zen2 the same as Zen+ and are we still talking about AM4 or is this AM4+? Originally the successor to Zen was called Zen+ in AMDs roadmaps. They later changed it to Zen2 but some people use that and Zen+ interchangeably. But now there is also a Zen refresh coming which is sometimes referred to as Zen+  Both should be compatable with with AM4.
|
|
#
?
Sep 5, 2017 17:27
|
|
|
I made a chart! It might be wrong! Hell yeah!
|
|
#
?
Sep 5, 2017 18:07
|
|
|
Motion to just use Zen, Zen Refresh, and Zen2?
|
|
#
?
Sep 5, 2017 18:15
|
|
|
new leak from wccftech
|
|
#
?
Sep 5, 2017 18:16
|
|
|
|
| # ? Apr 25, 2024 18:10 |
|
|
repiv posted:new leak from wccftech Big if true
|
|
#
?
Sep 5, 2017 18:48
|
|
