Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Stanley Pain
Jun 16, 2001

by Fluffdaddy

NewFatMike posted:

The E485 and E585 were just listed - I haven't looked into the older ones they're based off of to see if they have an open DIM.

Those are the two models I'm keeping an eye on.

Adbot
ADBOT LOVES YOU

NewFatMike
Jun 11, 2015

I wish they would have a higher resolution than 1080p. They'd essentially be ideal for me.

Yay for CAD! Finally figuring out what all the hubbub for screen real estate is about.

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE

FaustianQ posted:

Hm, I guess there is a reason for not clocking higher. Apparently die sizes are exactly the same, which indicates Pinnacle only gets optical improvements, it's not actually using improved design libraries probably for drop in compatibility for TR and EPYC; I can see Pinnacle EPYCs being very in demand. Might mean Raven Ridge will hit higher clocks as well since it'll never be an MCM design and therefore doesn't have to be concerned about it.

Lul that it's 2018 though and now, finally GloFo has a reasonable 14nm design. Maybe this'll fix the Polaris and Vega power issues for the refresh this year, like Polaris 10 @ 1450-1550Mhz @ 1.1v might consume 130W instead of 185-205W. Holy poo poo if GloFo has been that much of an Albatross.

ComputerBase:

quote:

As Joe Macri, AMD's Vice President & Chief Technology Officer, told ComputerBase, the manufacturer does not take full advantage of the new manufacturing process. AMD uses only the faster transistors and the better energy efficiency of 12LP. But not the smaller structures and thus the new libraries. That's why the Pinnacle Ridge die is the same size (213 mm˛) and has the same number of 4.8 billion transistors as Summit Ridge. Advantage: The development costs AMD has so significantly reduced.

shrike82
Jun 11, 2005

I got bitten in the rear end with the ThreadRipper/IOMMU issue for the first time - spent the entire weekend trying to debug why a dockerized tensorflow GPU model that worked on my office machine (7820X) wouldn't work on my home ThreadRipper setup.

Spent two days thinking it was a code issue without much success until I tried disabling IOMMU in Grub, worked perfectly.

Methylethylaldehyde
Oct 23, 2004

BAKA BAKA

shrike82 posted:

I got bitten in the rear end with the ThreadRipper/IOMMU issue for the first time - spent the entire weekend trying to debug why a dockerized tensorflow GPU model that worked on my office machine (7820X) wouldn't work on my home ThreadRipper setup.

Spent two days thinking it was a code issue without much success until I tried disabling IOMMU in Grub, worked perfectly.

Yeah, it's not exactly the most mature technology at the moment, but hopefully TR2 and the associated AGESA updates unfuck it. Or the Epyc AGESA updates tickle down over the next 6 months.

Malcolm XML
Aug 8, 2009

I always knew it would end like this.
Twitter analyst talk says that Intel is seriously loving up 10nm and hasn't back ported new architectures to 14nm so I'm not surprised a more modern arch like zen (vs skyline) is kicking rear end at a near 1ghz deficit.

GloFo 7nm is apparently delayed a bit but that's not a huge problem given that amd could dual source from TSMC

mdxi
Mar 13, 2006

to JERK OFF is to be close to GOD... only with SPURTING

FaustianQ posted:

Still think AMD needs some SKUs below 99$. Like, loving firesale the Bristol Ridge poo poo AMD, or get more Raven Ridge SKUs out. A 4C/4T, 6CU part and 4C/4T, 4CU part for 79$ and 59$ respectively, heck even the A12-9800 is worth it if it's only 50$.

AMD might not be doing it, but MicroCenter sure is:

NewFatMike
Jun 11, 2015

mdxi posted:

AMD might not be doing it, but MicroCenter sure is:


$80 for the 2200G has my heart pounding a little. That's some good poo poo.

buglord
Jul 31, 2010

Cheating at a raffle? I sentence you to 1 year in jail! No! Two years! Three! Four! Five years! Ah! Ah! Ah! Ah!

Buglord
So does this apparently CPU success increase resources for GPU R&D, or at least more CPU development? Since I got into PC gaming, intel/nvidia has always been the sole answer so its exciting to see that potentially change.

Craptacular!
Jul 9, 2001

Fuck the DH

Paul MaudDib posted:

Effectively, reviewers are testing maxed-out 2700Xs against stock 8700Ks. XFR2 pretty much extracts all the overclocking headroom right out of the box... which is good for lazy users who wouldn't otherwise bother with fiddling in the BIOS, but there is easily another 15-20% there on 8700Ks even above MCE speeds. Games still do care about clockrates, even if Ryzen has caught up on latency/etc.

It's like GPU Boost 3.0 on the 1070... it sucked up all the OC headroom, which looked really good against a reference 980 Ti that wasn't bothering to use its 30% OC headroom.

XFR2 isn't out of spec, though. PBO is. There's a lot to be said about getting a lot out of your chip without having to put it in danger plus void a warranty.

I only tried overclocking this 3770K chip a week ago, because I thought it's value had completely disappeared (haw haw, it's still selling for $250 good god almighty) and figured it was finally time to get my money's worth out of buying a K chip (which was only $10 or so back then) and an AIO cooler. I promptly created a whole bunch of WHEA errors that led to BSODs in Overwatch even though the chip was passing stability tests, and gave up and went back to stock because there's often a correlation between WHEA errors and chip degradation.

Some people don't want to risk that, and they especially don't want to risk it on hardware that's new. I never OCed a CPU until this one because I used to have to change my CPU every two years in the Pentium era anyway and it never made sense to overclock an old one versus buying the new hardware you'll need eventually. Also in those days new hardware had immediate noticeable benefits, and we've gone from "IE4 loads in under eight seconds now holy poo poo" to "I can't see any differences between my 8700K and my 4790K."

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!

So is this in agreement or am I misinterpreting things? Either way, past 14nm products should have no issue getting the benefits of the new "12nm process" with zero effort.

mdxi posted:

AMD might not be doing it, but MicroCenter sure is:


Eh, 49$ for a A8-9600 is still kind a steep, should be 39$ IMHO, for 30$ more the R3 2200G is just going to demolish it, and the R5 1400 is also nutso value if you already have a dGPU.

NewFatMike posted:

$80 for the 2200G has my heart pounding a little. That's some good poo poo.

Basically murders anything in raw value, @ 99$ it's real good but 79$ not even Intel's low end stuff could reasonably compete.

Malcolm XML posted:

Twitter analyst talk says that Intel is seriously loving up 10nm and hasn't back ported new architectures to 14nm so I'm not surprised a more modern arch like zen (vs skyline) is kicking rear end at a near 1ghz deficit.

GloFo 7nm is apparently delayed a bit but that's not a huge problem given that amd could dual source from TSMC

So basically 8C Coffeelake vs Ryzen 3700X in 2019, lmao. Holy poo poo Intel is going to get murdered.

buglord posted:

So does this apparently CPU success increase resources for GPU R&D, or at least more CPU development? Since I got into PC gaming, intel/nvidia has always been the sole answer so its exciting to see that potentially change.

Yes, sales are strong enough AMD is solidly in the black in nearly a decade and has already been on a massive hiring spree and moving resources around. If Intel can't get their poo poo together for 10nm in 2019, AMD will retake the performance crown. Nvidia is a different matter entirely, and the best you can hope for AMD being competitive at all points in the stack is like, late 2020 at best, 2021 otherwise.

SwissArmyDruid
Feb 14, 2014

by sebmojo

Happy_Misanthrope posted:

This bodes extremely well for the next consoles in 2-3 years too.

Iiiiiiiiiii hesitate to agree. Consoles are basically an APU, and if AMD can't rise to the task on the GPU side of the equation, what with Vega being an unmitigated disaster except at low core counts, and Navi not arriving before 2019-2020, it's still not looking hot.

Palladium
May 8, 2012

Very Good
✔️✔️✔️✔️

Craptacular! posted:

XFR2 isn't out of spec, though. PBO is. There's a lot to be said about getting a lot out of your chip without having to put it in danger plus void a warranty.

I only tried overclocking this 3770K chip a week ago, because I thought it's value had completely disappeared (haw haw, it's still selling for $250 good god almighty) and figured it was finally time to get my money's worth out of buying a K chip (which was only $10 or so back then) and an AIO cooler. I promptly created a whole bunch of WHEA errors that led to BSODs in Overwatch even though the chip was passing stability tests, and gave up and went back to stock because there's often a correlation between WHEA errors and chip degradation.

Some people don't want to risk that, and they especially don't want to risk it on hardware that's new. I never OCed a CPU until this one because I used to have to change my CPU every two years in the Pentium era anyway and it never made sense to overclock an old one versus buying the new hardware you'll need eventually. Also in those days new hardware had immediate noticeable benefits, and we've gone from "IE4 loads in under eight seconds now holy poo poo" to "I can't see any differences between my 8700K and my 4790K."

It's a crime not to take advantage of the NA Ebay CPU market who seriously are as dumb as bricks, where over here I will be lucky to sell my 4790K for $150 because people actually pays attention to cost-benefit ratios. Also good luck anyone finding P/Z S1155 mobos that aren't degraded by this point.

And yeah I definitely will be stocking more 16/32GB DDR4 for future AMD poo poo if we get 2016 prices ever again.

SwissArmyDruid posted:

Iiiiiiiiiii hesitate to agree. Consoles are basically an APU, and if AMD can't rise to the task on the GPU side of the equation, what with Vega being an unmitigated disaster except at low core counts, and Navi not arriving before 2019-2020, it's still not looking hot.

My wild guess is NV is probably talking with Samsung's custom ARM division for the same reasons.

Palladium fucked around with this message at 03:56 on Apr 23, 2018

SwissArmyDruid
Feb 14, 2014

by sebmojo

Palladium posted:

My wild guess is NV is probably talking with Samsung's custom ARM division for the same reasons.

That may be true, but based on how the X.B.O.X. only came out at the end of last year, it probably means they aren't going to launch a new whole-step console any time before 2020, or wind up pissing off people who had just bought the new half-step console.

It makes sense for Sony to just sit there and print money, without the pressure to put out a new console any time before then.

I just don't see Sony stepping away from the x86 architecture. The benefits of a common ISA with PC, let alone PC and XBOX for developers, are significant.

SwissArmyDruid fucked around with this message at 04:54 on Apr 23, 2018

buglord
Jul 31, 2010

Cheating at a raffle? I sentence you to 1 year in jail! No! Two years! Three! Four! Five years! Ah! Ah! Ah! Ah!

Buglord

Holy crap the more you know.

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE

FaustianQ posted:

So is this in agreement or am I misinterpreting things? Either way, past 14nm products should have no issue getting the benefits of the new "12nm process" with zero effort.

Correct, just sourcing your claim here.

FaustianQ posted:

So basically 8C Coffeelake vs Ryzen 3700X in 2019, lmao. Holy poo poo Intel is going to get murdered.


Yes, sales are strong enough AMD is solidly in the black in nearly a decade and has already been on a massive hiring spree and moving resources around. If Intel can't get their poo poo together for 10nm in 2019, AMD will retake the performance crown. Nvidia is a different matter entirely, and the best you can hope for AMD being competitive at all points in the stack is like, late 2020 at best, 2021 otherwise.

Holy poo poo yes Intel is so turbofucked, if 7nm comes remotely close to what's promised and Intel doesn't make a massive leap above what they've shown to be able to do, they are so hosed.

I'm pretty much holding off upgrading my HEDT desktop to see what happens next year. 5960X for $360 or 6950X for $640 is appealing but the showdown next year is probably gonna be real good.

8-core vs 5 GHz: why not both? (and Zen2 may actually be a larger CCX too)

Klyith
Aug 3, 2007

GBS Pledge Week
consoles have a huge lead-time between nailing the specs and putting out the box, because they have to give devs time to make games for the launch. so they're definitely not using navi. and they're gonna be using AMD graphics because that's the only game in town -- the conditions that drove both consoles to use x86 & AMD have not materially changed. so the PS5 at least (which apparently is already in super-early dev kit land) will probably use ryzen & vega derived architecture.

consoles don't care that vega is a trash fire, they're not really competing against PCs.

however, here's some reasons vega might not be quite as bad a trash fire as you think:
1. it was priced for it's performance competition in the crypto market, not the video games market. AMD has sold big GPUs for minimal margin before, so Vega probably could have been priced cheaper. they didn't because crypto was distorting everything.
2. therefore the only people who bought them were crypto-miners (idiots) and AMD super-fans (also idiots), with the vast majority of purchases being crypto idiots. Vega doesn't even show up on steam hardware charts.
3. given that, zero work is being done to optimize games for vega. Vega really did have arch improvements for GCN, but I doubt they're really getting any attention.
4. GCN is old and creaky, but it's gotten a hell of a lot of lifespan extension already and a lot of targeted optimization based on the consoles. If the next consoles use Vega, that stuff that's being ignored might actually get used!

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE

NewFatMike posted:

$80 for the 2200G has my heart pounding a little. That's some good poo poo.

Are there any mITX boards that are decent? If I make the trek out to Microcenter (about 1h) if I could grab a RR box for $200 or less (ideally $150 or less). Probably not going to go out there for just that (it's about $15 in gas just for the trip) but I'd probably do it the next time I was out there. Right now my TV PC does not have H265 decode... and RR does not need a discrete GPU for that, so I wouldn't need to rebuild that PC, I could build a MBox M350 instead.

From what I've heard a lot of the mITX boards have terrible 3+2 VRMs and other dumb decisions though.

Paul MaudDib fucked around with this message at 06:40 on Apr 23, 2018

Craptacular!
Jul 9, 2001

Fuck the DH
There's not a lot of ITX boards. Your choice is to either buy the Strix now with all it's bells and whistles, or wait for the MSI board that's been given press photos, reviews, and a cameo at CES but still isn't at Amazon or Newegg. But given that it's been in the pipe since January who knows if anyone will ever be able to buy one.

Gigabyte and ASRock have boards, but if you're peculiar about VRMs it's either Strix or MSI Vaporware.

Craptacular! fucked around with this message at 08:00 on Apr 23, 2018

Pablo Bluth
Sep 7, 2007

I've made a huge mistake.
Yet another bios update for my board seems to have doubled the number of DOCP profiles in the bios, and I can now finally run the memory at the rated 3000 instead of having to settle for 2933...

Anarchist Mae
Nov 5, 2009

by Reene
Lipstick Apathy
Amazingly there are a lot of options already available locally. I'd like to upgrade from my B350 board to something more likely able to handle overclocking my 1700, and I also need something with 8 SATA ports for more HDDs. Sadly it's all expensive as gently caress right now, 50000huf is about $200USD, and the first motherboard that fits my SATA requirements is $250USD.

Edit: I just found that virtualisation appears to be broken now, I think it's due to a bios update:

https://www.youtube.com/watch?v=ov17CR4tTwE

It could just be a VirtualBox issue, I hope it is as that means I can probably get someone to fix it. If it's an ASRock issue then I'm hosed since I've had an open support issue with them for two months and not heard a drat thing.

Here's info I got from the one time the Fedora machine booted:


Anarchist Mae fucked around with this message at 11:41 on Apr 23, 2018

ufarn
May 30, 2009
If I get a 2700X with 3200 DDR4, and X470 motherboard, how many of these will be reusable for Zen 2? I'm assuming AM4 will still support Zen 2, as crazy as that sounds on its face.

Then I could get a Zen 2+ in two-three years as an upgrade and then perhaps get a dedicated motherboard with better performance down the line.

sauer kraut
Oct 2, 2004

Paul MaudDib posted:

Holy poo poo yes Intel is so turbofucked, if 7nm comes remotely close to what's promised and Intel doesn't make a massive leap above what they've shown to be able to do, they are so hosed.

I wouldn't dance on Intels grave yet until we actually see a finished 7nm high power GloFo product.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Measly Twerp posted:

It could just be a VirtualBox issue
So uh... Why VirtualBox? Especially if you're virtualizing other Linux instances? With KVM and the related paravirt stuff in the kernel, on either side of the VM, would surely make things smoother and more efficient?

ufarn posted:

Then I could get a Zen 2+ in two-three years as an upgrade and then perhaps get a dedicated motherboard with better performance down the line.
Will there actually be a Zen 2+? I thought the plus thing was just with Zen 1, per their roadmap? After Zen 2, it goes straight to 3, which is supposed to come with DDR5 and PCIe5 and therefore a new socket.

--edit: Nevermind, outdated roadmap.

Combat Pretzel fucked around with this message at 14:20 on Apr 23, 2018

Klyith
Aug 3, 2007

GBS Pledge Week

ufarn posted:

If I get a 2700X with 3200 DDR4, and X470 motherboard, how many of these will be reusable for Zen 2?

if you believe AMD promises, all of it. you will just need to update your bios and drop in a zen 2 chip.


sauer kraut posted:

I wouldn't dance on Intels grave yet until we actually see a finished 7nm high power GloFo product.

until I see a working 7nm production line somewhere on the planet, I'm not putting money on anyone.

Arzachel
May 12, 2012

sauer kraut posted:

I wouldn't dance on Intels grave yet until we actually see a finished 7nm high power GloFo product.

I'm more than a little skeptical of GloFlo's 7nm claims but at least AMD isn't locked into them anymore and can dual source chips from TSMC.

buglord
Jul 31, 2010

Cheating at a raffle? I sentence you to 1 year in jail! No! Two years! Three! Four! Five years! Ah! Ah! Ah! Ah!

Buglord

Unrelated but I really enjoy your avatar.

GRINDCORE MEGGIDO
Feb 28, 1985


8 core coffeelake (autocorrected to coffeemaker) at around 5ghz is..... Going to be quite potent.

I'd be 100% impressed if AMD even caught up to that in IPC or clocks in 2019.

GRINDCORE MEGGIDO fucked around with this message at 18:57 on Apr 23, 2018

Khorne
May 1, 2002

Combat Pretzel posted:

So uh... Why VirtualBox? Especially if you're virtualizing other Linux instances? With KVM and the related paravirt stuff in the kernel, on either side of the VM, would surely make things smoother and more efficient?

Will there actually be a Zen 2+? I thought the plus thing was just with Zen 1, per their roadmap? After Zen 2, it goes straight to 3, which is supposed to come with DDR5 and PCIe5 and therefore a new socket.

--edit: Nevermind, outdated roadmap.
New PCIE shouldn't require a new socket. It just won't be supported on older motherboards like in the past.

DDR5 will, obviously.

Khorne fucked around with this message at 16:25 on Apr 23, 2018

Anarchist Mae
Nov 5, 2009

by Reene
Lipstick Apathy

Combat Pretzel posted:

So uh... Why VirtualBox? Especially if you're virtualizing other Linux instances? With KVM and the related paravirt stuff in the kernel, on either side of the VM, would surely make things smoother and more efficient?

Because at work we use vagrant to share virtual machine configurations to run on OS X or Linux.

NewFatMike
Jun 11, 2015

buglord posted:

Unrelated but I really enjoy your avatar.

I've been thinking the same thing.

35W TDP R3 2200GE and R5 2400GE just showed up. No bundled heatsink noted, but I wonder if these will find themselves in mobile workstations like the HQ processors from Intel.

https://techreport.com/news/33547/amd-ryzen-5-2400ge-and-ryzen-3-2200ge-pop-up-with-35-w-tdps

Eletriarnation
Apr 6, 2005

People don't appreciate the substance of things...
objects in space.


Oven Wrangler
Comparing the 2400GE to a 2700U, I'm having a hard time seeing why you would use the former in a laptop unless it's substantially cheaper. Presumably the configurable TDP would allow a 2700U to turbo up to similar performance if set to 25W, and all the specs look the same except for the Vega IGP having 10 CUs in the 2700U vs. 11 in the 2400GE. I definitely expect to see the 2400GE in AIOs or SFF desktops though.

e: I'm actually a little surprised that AMD isn't teasing versions of 6-8 core Ryzen for mobile workstations. My guess would be that they either don't think there's sufficient demand or that they are having difficulty producing a sufficiently compact version of the package, but I don't really have any idea.

Eletriarnation fucked around with this message at 16:49 on Apr 23, 2018

Llamadeus
Dec 20, 2005

Eletriarnation posted:

Comparing the 2400GE to a 2700U, I'm having a hard time seeing why you would use the former in a laptop unless it's substantially cheaper.
You wouldn't, but they do two variants the same way Intel does: same silicon, one is a BGA laptop chip and the other is for socketed desktops, eg 7700HQ vs 7700T.

NewFatMike
Jun 11, 2015

Eletriarnation posted:

e: I'm actually a little surprised that AMD isn't teasing versions of 6-8 core Ryzen for mobile workstations. My guess would be that they either don't think there's sufficient demand or that they are having difficulty producing a sufficiently compact version of the package, but I don't really have any idea.

There probably isn't enough demand on the high end to justify a new piece of silicon that wouldn't be addressed by the i7-G market (and AMD still get their money from that without spending any marketing money*).

*Given AMD's historical marketing campaigns, this is probably for the best.

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!
Well Pinnacle is an even better low power solution so maye now?

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!
Regarding B-die memory, what are the chances I will be able to push it a lot, regardless of what the sticks say (e.g. pushing a DDR4-2400 stick to 3200)? Mostly interested about this to get DDR4-3200 ECC sticks.

Measly Twerp posted:

Because at work we use vagrant to share virtual machine configurations to run on OS X or Linux.
It apparently supports KVM, too?

Combat Pretzel fucked around with this message at 18:46 on Apr 23, 2018

Anarchist Mae
Nov 5, 2009

by Reene
Lipstick Apathy

Combat Pretzel posted:

Regarding B-die memory, what are the chances I will be able to push it a lot, regardless of what the sticks say (e.g. pushing a DDR4-2400 stick to 3200)? Mostly interested about this to get DDR4-3200 ECC sticks.

It apparently supports KVM, too?

OS X does not support KVM. There's nothing wrong with VirtualBox, it meets all our needs and it's highly likely that KVM would have the exact same problem.

isndl
May 2, 2012
I WON A CONTEST IN TG AND ALL I GOT WAS THIS CUSTOM TITLE

NewFatMike posted:

*Given AMD's historical marketing campaigns, this is probably for the best.

I for one am waiting for my monster truck laptop to show up any day now. :colbert:

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Measly Twerp posted:

OS X does not support KVM. There's nothing wrong with VirtualBox, it meets all our needs and it's highly likely that KVM would have the exact same problem.
Ah, I had the impression that Vagrant was supposed to translate between the virtual machine formats.

As far as KVM does, the developers are very diligent when it comes to figuring out workarounds for odd issues like that, assuming you're staying up to date with the kernel. Last I cared about VirtualBox, it didn't seem like much effort is spent on dealing with hardware oddities.

Adbot
ADBOT LOVES YOU

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!

isndl posted:

I for one am waiting for my monster truck laptop to show up any day now. :colbert:

What would 35W Threadripper even look like? 1.5Ghz base, 2.8Ghz boost?

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply