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
wargames
Mar 16, 2008

official yospos cat censor

SourKraut posted:

If I can get a R5 1600X for $217, is that a good "jump in" price? It seems like a lot of people are going with the R7 1700 (non-X), so want to make sure I'm not missing something.

220 for a 1600x is pretty good, and could get to 3.9 oc.

Adbot
ADBOT LOVES YOU

wargames
Mar 16, 2008

official yospos cat censor

Palladium posted:

Much of technical advice out on teh Internets can be summed as "my toys are more expensive than yours".

or "it works for me I don't know why it doesn't for you"

wargames
Mar 16, 2008

official yospos cat censor

Drakhoran posted:

Since we're linking to Videocardz.com, anyone feel like a 48 core "Starship" chip on 7nm next year?

That would be awesome I wonder what the pricing will be or if its server only cpu.

wargames
Mar 16, 2008

official yospos cat censor

SwissArmyDruid posted:

Funny you should mention that. We finally have a Threadripper, and it's a codename for Ryzen 9 parts that correspond to Intel -E/-X parts.

Usual WCCFT caveats apply: Salt now so you're not salty later.

http://wccftech.com/amd-ryzen-9-lineup-threadripper/

Of note:



I can't be the only one who immediately thought "K6!" as soon as I saw that 1998 number, right?

But the most important question goons have what about ecc?

wargames
Mar 16, 2008

official yospos cat censor

MrBadidea posted:

My old buildserver finally started to choke a few weeks ago, so I bit the bullet and replaced it before it died entirely. I gotta throw some more memory at it, and actually sort out the memory clocks (the current stuff absolutely refuses to post above 2133), but it's absolutely smoking what I'd have been able to afford for the same budget from Intel.

I finally got down to getting some actual comparison numbers yesterday using UE4 lighting builds. Something something non-scientific, single runs, same map, same quality settings. Default settings uses 6 threads for the Intel machine and 14 for the Ryzen. The 1700X machine was paging hard for most of it too, so I have no doubt another 16Gb would make this comparison even more ridiculous.


Actual compile times for the engine got similarly destroyed; from ~90 minutes down to an average of ~20 for a fully clean build :aaaaa:

Impressions so far are pretty good. I've not had a chance to sit down and game on it yet to get a feel for it, but for sure it destroys heavy workloads at a fraction of the price. The only thing letting it down so far has been bios related; it'll occasionally start up and not detect any disks until it gets rebooted again, and it's strangely reluctant to run the memory any faster than 2133.

Post like this make me want to get a zen machine up and find a cheap colo place to keep it at.

wargames
Mar 16, 2008

official yospos cat censor
isn't the 1050s built by samsung not tmsc?

wargames
Mar 16, 2008

official yospos cat censor

Risky Bisquick posted:

mITX x370 instead of the x300 :raise: x300 seem to be vaporware

x300 is the super low end right so maybe for partners only that do enterprise poo poo boxes?

wargames
Mar 16, 2008

official yospos cat censor

Drakhoran posted:

No, prices on all kinds of cryptocurrency are extremely volatile. One week miners will be buying up every card making them hard to get for actual gamers, next week they sell all their used cards on eBay bringing retail prices crashing down.



In unrelated news. Here is a picture of two Asrock X399 boards. Looks like the official name for the Threadripper socket will be TR4.

The people who claim a single pea sized dab of thermal grease will have something break in their mind once that lands in front of them.

wargames
Mar 16, 2008

official yospos cat censor

eames posted:

... and terrible performance because of virtualization bugs that AMD won't acknowledge. :(

https://community.amd.com/thread/215931

The promised fix (AGESA 1006) doesn't help with NPT performance, it only fixes groupings to enable GPU passthrough.

They know of immou groups issues and are actively working on it.

wargames
Mar 16, 2008

official yospos cat censor

eames posted:

We'll find out soon enough I guess.

Speaking of Threadripper, Linus dropped some pretty heavy hints that Threadripper is two Ryzen 7 cores on a single package. I always knew that the architecture would be modular but having what appears to be two identical sockets merged into one was eye opening. Hopefully performance won't suffer too much if it's an SMP-like architecture.

https://www.youtube.com/watch?v=nynBEGCtg90

If the interposer is pretty expandable it shouldn't be an issue, and amount to huge savings.

wargames
Mar 16, 2008

official yospos cat censor

Methylethylaldehyde posted:

If they can get the infinity fabric fast enough and low enough latency, then it shouldn't be too much of an issue, as long as they address the NUMA issues and make sure that all the context switching costs are known/available to the OS. It would be a great way to salvage marginally performing chips, or chips that are super great except for that huge defect in core 3.

Since its single cpu NUMA doesn't come into play with infinity fabric, do note that 1700-1800x parts don't do numa and had other fixes come in without having to use NUMA as a scheduler.

wargames
Mar 16, 2008

official yospos cat censor

Scarecow posted:

So if they are doing price cuts to the top end of the r7s does that mean threadripper could be coming in at crazy prices vs intels offerings?

A brick of gold is coming in at a crazy price vs what intel offers.

wargames
Mar 16, 2008

official yospos cat censor

Scarecow posted:

Maybe the rumors of amd getting crazy high yeilds is true then

This was the hope with zen smaller dies on an interposer for mondo yields.

wargames
Mar 16, 2008

official yospos cat censor
If this is true, zen is even more disruptive then we thought and amd might be profitable one day.

https://twitter.com/BitsAndChipsEng/status/870621722051063808

https://twitter.com/BitsAndChipsEng/status/870597420958326788

wargames fucked around with this message at 15:41 on Jun 2, 2017

wargames
Mar 16, 2008

official yospos cat censor
Vega is the new bulldozer.

wargames
Mar 16, 2008

official yospos cat censor

FaustianQ posted:

There is no way Threadripper and Eypc don't get picked up for enterprise and worskatation use at these prices, it's literally AMD CPU+MB ≤ Intel CPU. Knowing AMD, they'd throw in discounts for buying completely from them, so that means more GPU sales and more sales of their rebranded SSDs and memory. Has AMD made any moves to either source Qualcomm or acquire their own for ethernet controllers?

The HBCC memory controller AMD is talking about has me thinking, could AMD turn HBM2 into a nonvolatile storage medium, or even a PCIE Ramdisk?

Doubtful amd is going into that line of research but look at what intel is doing with cross point

https://www.youtube.com/watch?v=GsmoHsB6ITc

wargames
Mar 16, 2008

official yospos cat censor

VulgarandStupid posted:

These motherboard chipset conventions are getting a little ridiculous. Intel has the X299 chipset, so we'll call our competing chipset the X399. That will show them.

AMD did steal the FE short hand from Nvidia so everyone steals everything.

wargames
Mar 16, 2008

official yospos cat censor
Amd needed to make a great and they did, They also needed intel to gently caress up hard and that is happening?

wargames
Mar 16, 2008

official yospos cat censor

Comatoast posted:

What in the world? It's like a 23 minute long MacRumors video.

Guy is a fanboy but some of his thoughts aren't completely trash.

wargames
Mar 16, 2008

official yospos cat censor

Kazinsal posted:

I wonder if there are some chips that really lost the silicon lottery and just don't have enough volts at stock clocks to keep up under a really intense long-running 100% load situation.

Would zen2 with just a slight bump to volts all around be a lazy fix for such things or is zen 2 going to be on 7nm?

Also what happened to intels 10nm process?

wargames
Mar 16, 2008

official yospos cat censor

Kazinsal posted:

I suspect a die shrink would be in the cards at some point if AMD can end friendship with GloFo, now Samsung is my best friend. There's no way they could do a competitive 10nm part on the GloFo process.

Cannonlake (Intel's Kaby Lake die shrink) has been pushed back a bit last I checked.

AMD is planning on skipping 10nm and using ibm/samsung/gloflo 7nm in 2018 for zen2, and zen3 will use 7nm refined.

wargames
Mar 16, 2008

official yospos cat censor

Measly Twerp posted:



I can finally afford to upgrade. In three days I'm going to put my i3-6100 in a box and forget it ever existed.

What does 134,699 feet convert to real money?

wargames
Mar 16, 2008

official yospos cat censor

Dante80 posted:



That is $592. Biggest online retailer here in Greece.

gently caress my country. That is why I bought most of the hardware for my 1600X rig from ze Germans.



Add to that about 30€ for shipping to Greece, the total comes to $488.

Would it be cheaper to pick it up from germany since 30 euros to ship is kind of insane?

wargames
Mar 16, 2008

official yospos cat censor

A SWEATY FATBEARD posted:

R7 1700 is $422 in Croatia. ASROCK AB350 Pro4 board is $135. I'm obviously itching to retire my FX-8350 rig, yeah, I know, don't ask. :ughh:

I'm so sorry, hell this bundle with 16gb of ram on newegg is 500

https://www.newegg.com/Product/ComboBundleDetails.aspx?ItemList=Combo.3406453

wargames
Mar 16, 2008

official yospos cat censor

Paul MaudDib posted:

Well, that's literally the design AMD proposed for their Exascale Heterogeneous Processor concept, which IIRC was in response to some early-stage/2020-delivery RFP from one of the National Labs or something. And HBM2 lets you go up to 8-high per stack so conceptually you could easily do at least 32 GB of VRAM.

You're right about the challenge posed by heat though, that's a datacenter chip, not something you would run in your PC. You would likely want to run that chip under liquid cooling (iirc NVIDIA popped the cherry on this a while back and their Tesla rack servers use liquid cooling)

Not like an AIO is all that expensive these days, though. $100 gets you a 240mm AIO that can easily do 500W+ given a sufficiently big coldplate/heatspreader (see: R9 295X2 with its 120mm AIO).

Motherboard on chip?

wargames
Mar 16, 2008

official yospos cat censor

Wirth1000 posted:

Ok, I don't know what to do with this build now. Somebody suggest benchmarks or something. 1600X + B350 + 16GB RAM

Only thing that comes to mind to test cpu is Cities skylines but I don't know how multi treaded that is.

wargames
Mar 16, 2008

official yospos cat censor

VealCutlet posted:

Got my 1600 to 4GHz tonight for some benchmarks, waiting patiently for threadripper... if it makes it to 5GHz I would probably jizz all over my mobo. http://www.3dmark.com/fs/12889394

Threadripper will not make it to 5gz, Threadripper 2 on Zen 2 might. But current threadripper on zen 1 will boost as much as your 1600.

wargames
Mar 16, 2008

official yospos cat censor

Kazinsal posted:

They ran the Intel systems with the RAM downclocked to 2133 MHz too? WTF?

Slower rams effect intel way less then amd.

wargames
Mar 16, 2008

official yospos cat censor

Combat Pretzel posted:

GPU performance is currently irrelevant for cloud providers.

If Microsoft wants to compete with amazon when it comes to tensor computer nodes they will.

wargames
Mar 16, 2008

official yospos cat censor

Mr.Radar posted:

Ryzen = Desktop
Threadripper = Workstation
Epyc = Server

We need to rearrange that

Ryzen = Desktop
Epyc = Server
Kittyzen = Mobile
Threadripper = Workstation

so AMD can REKT cpus.

wargames
Mar 16, 2008

official yospos cat censor

Rastor posted:

Here's a Gigabyte mini-ITX board for the AMD Ryzen socket AM4 platfrom






now one for THREADRIPPER.

wargames
Mar 16, 2008

official yospos cat censor

Palladium posted:

I think most of us forgot laptop CPUs constitutes maybe like 70% of Intel's consumer market and AMD has been non existent there for a decade.

ultrabay2000 posted:

I wouldn't buy one but they seem to do alright in the Walmart laptop segment. Seems like a large chunk of cheap laptops I see have AMD CPUs.

as one of those people that got a 300 dollar best buy laptop with an A8 amd processor in it, its not bad the only thing I hated about the laptop was its harddrive but I replaced it and it does it job just fine.

wargames
Mar 16, 2008

official yospos cat censor

Paul MaudDib posted:

I see AMD is embracing Intel's nonsense model names.

At least it follows a "higher numbers are faster or have more cores" rule.

Intel has gone to gold, silver, platinum for their xeons so good on amd for atleast sticking to numbers

wargames
Mar 16, 2008

official yospos cat censor

Nam Taf posted:

This triggers me because it makes sense until the 1950/X which should be the 1960/X

It'd then follow the 19x0 where x is the 2nd digit of the core count. I mean it could be far better but at least it'd make sense.

Right now it's so close but then goes full AMD :allears:

Edit: missed the 1930X loving everything up.

what about the 1910? basically its all bad names.

wargames
Mar 16, 2008

official yospos cat censor

Munkeymon posted:

That's just the L3 cache :getin:

But seriously I didn't realize HBM was slower than regular DDR4. Huh!

Current HBM2 on vega(poo poo card) is 900ish mhz but drops to 500 if you change anything.

wargames
Mar 16, 2008

official yospos cat censor

Cygni posted:

GPUs are theoretically more suited to multiple die packages too due to the parallel workload. Epyc's latency for going between the dies is fairly atrocious, and AMDs way of addressing that was essentially working around that reality (just like Intel did with the Pentium D). GPUs wouldn't deal with that quite as much. I mean heck, the concept you're describing for Navi is basically VSA-100. :v:

The problem Navi faces is the same one AMD has faced for years: Nvidia isn't just sitting around waiting for them to catch up. Volta is already at the reticle limit of a 12nm process, so I imagine Nvidia is as fully aware of the manufacturing constraints and future performance needs as AMD is.

but navi i think is planed for 7nm which was dev'd by ibm/samsung/gloflow.

wargames
Mar 16, 2008

official yospos cat censor
AMD GPU needs a blank sheet for navi and forget Hawaii ever existed.

wargames
Mar 16, 2008

official yospos cat censor

FaustianQ posted:

AMD's execution comes down to a sharp disconnect between marketing and engineering departments, lovely marketing, and not enough money for marketing and engineering combined. If Ryzen and Epyc do well you'll first see the changes in marketing, and then maybe in 2019-2021 for products.

Also apparently AMD is going on a hiring bonanza for marketing and engineering positions, so uh, guess things are in fact going well?

with stock prices up and ryzen generating enough cash to keep the debters off their back, turn the rest of the cash into much needed company investments.

wargames
Mar 16, 2008

official yospos cat censor

priznat posted:

I'm curious if the current version of InfinityFabric phy can support PCIe Gen4 speeds (8GHz, 16GT/s), because it would be pretty impressive if they could jump right into the Gen4 ecosystem. There's a few new features they'd have to implement to get to the Gen4 spec of course.

Have there been any blurbs on how fast the IF links can go? I've seen things like "512Gps for GPUs" but it doesn't say the width.

It doesn't need to, remember AM4 is suppose to be supported for 5 years.

Adbot
ADBOT LOVES YOU

wargames
Mar 16, 2008

official yospos cat censor

Combat Pretzel posted:

So uh, if it's super high bandwidth, why does it get hosed over by the memory clock? Sounds like its data transfers are gated by the memory controller? If so, why?! I guess I kind of fail to see why it gets punished so hard versus Intel's ring bus.

Because IF is half the speed of ram this is a hard setting in Zen1 so if you increase ram speed you increase IF.

Combat Pretzel posted:

Oh, I have things to occupy all 16 cores once a while. What I'm more concerned about is when there will be situations when NUMA is going to be a problem, i.e. mismatch between core a thread is running on and where memory allocations are. At a certain point depending on memory pressure, allocations will start to cross NUMA memory regions. If the IF bandwidth between CCX pairs is indeed so much worse as speculated, this is going to be noticeable in some form.

On the other hand, if IF is 32 bytes wide and can run an PCIe speeds, as said a couple of posts above, why the hell are there even bandwidth issues that depend on RAM speed?

Also Zen1 doesn't use Numa for CCX talk it uses a new scheduler that microsoft and AMD came up with to account for the slight higher CCX talk.

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