New Zealand can eat me posted:I have figured out how to get my 2x16 DDR4-3000 CL15 to run at 2666 with DOCP enabled. Just enable it then manually set the speed from 3000 to 2666. What I don't understand is why the bios detects 15C but RYZEN MASTER is setting it to 16. I read a random amazon review that said Ryzen supposedly prefers even numbered CL timings, and had luck running their 3200 kit at CL14 instead of CL16. I have not had any success with that and attempting to do so results in me having to reset CMOS and start anew. I'm going to try the 4x8 kit again with this month's bios and see if that's good for anything tomorrow. Go watch this video on Ryzen RAM OCing. Short version: The Ryzen memory controller is forcing 1T command rate, you need RAM using Samsung B die chips if you want to get decent timings on Ryzen, also if you want anything over 2666 your motherboard needs an external clock generator, only some of the high end X370 boards have that so 2666 is as high as you get on the ASUS PRIME B350 PLUS.
|
|
#
?
Mar 25, 2017 14:37
|
|
|
|
| # ? May 21, 2024 11:54 |
|
|
all this trouble for the least significant factor in your computer's performance
|
|
#
?
Mar 25, 2017 14:41
|
|
Potato Salad posted:all this trouble for the least significant factor in your computer's performance Memory data rate has a very noticeable effect on FPS in some games, GTAV and TW3 being good examples where you can gain 10-15% FPS by going from DDR4-2400 to 3200. Also it's gonna matter a lot for the datacenter versions of the chips where memory performance really matters.
|
|
|
#
?
Mar 25, 2017 14:56
|
|
|
Potato Salad posted:all this trouble for the least significant factor in your computer's performance AMD's infinity fabric runs at half the speed of your ram. Cranking that up as much as possible helps out with the bottleneck in there.
|
|
#
?
Mar 25, 2017 15:42
|
|
|
AVeryLargeRadish posted:Go watch this video on Ryzen RAM OCing. Short version: The Ryzen memory controller is forcing 1T command rate, you need RAM using Samsung B die chips if you want to get decent timings on Ryzen, also if you want anything over 2666 your motherboard needs an external clock generator, only some of the high end X370 boards have that so 2666 is as high as you get on the ASUS PRIME B350 PLUS. That doesn't make any sense because it shows up as 2T sometimes!!! Also they literally list speeds up to 3200mhz in the specifications, and there are at least a few sets of ram out there that will DOCP 3200mhz, iirc one is for sure listed in the QVL. Maxwell Adams posted:AMD's infinity fabric runs at half the speed of your ram. Cranking that up as much as possible helps out with the bottleneck in there. Came here to say this. Potato Salad is wrong, and should feel bad.
|
|
#
?
Mar 25, 2017 15:59
|
|
New Zealand can eat me posted:That doesn't make any sense because it shows up as 2T sometimes!!! Also they literally list speeds up to 3200mhz in the specifications, and there are at least a few sets of ram out there that will DOCP 3200mhz, iirc one is for sure listed in the QVL. I'm not saying that the Ryzen controller always forces 1T command rate, it just does that very often and many memory kits can't deal with that at higher speeds. And yeah, there are memory sets with DOCP profiles for high speeds, but they depend on a modified base clock to work, like the G-Skill kits need to run at around a 120MHz base clock to actually hit 3200 and to change the base clock you need an external clock generator which only some motherboards have, all of this was covered in the video.
|
|
|
#
?
Mar 25, 2017 16:11
|
|
|
No but I'm talking about the QVL for my specific motherboard. That has confirmed working DOCP3200 profiles.
|
|
#
?
Mar 25, 2017 16:15
|
|
|
Potato Salad posted:all this trouble for the least significant factor in your computer's performance It has significant positive effects on minimum fps
|
|
#
?
Mar 25, 2017 16:20
|
|
New Zealand can eat me posted:No but I'm talking about the QVL for my specific motherboard. That has confirmed working DOCP3200 profiles. Does your RAM come with those profiles and have they been specifically tested with your model of motherboard? I'm only seeing two kits, the Galaxy HOF4CALCS3600K17LD162C and the Corsair CMK16GX4M2B3600C18, both DDR4-3600 kits that are approved for your board with a Ryzen CPU at 3200MHz.
|
|
|
#
?
Mar 25, 2017 20:16
|
|
|
AVeryLargeRadish posted:Does your RAM come with those profiles and have they been specifically tested with your model of motherboard? I'm only seeing two kits, the Galaxy HOF4CALCS3600K17LD162C and the Corsair CMK16GX4M2B3600C18, both DDR4-3600 kits that are approved for your board with a Ryzen CPU at 3200MHz. Those do, yes, which invalidates everything you said above, which was primarily the point I was making. Regardless, we should see another bios patch before April 9th that will change all of this anyways.
|
|
#
?
Mar 25, 2017 20:24
|
|
|
Am I doing this right? 
|
|
#
?
Mar 26, 2017 04:34
|
|
|
The CPU-Z benchmark is inaccurate for Ryzen, as it was made with 256MB L3 cache in mind and the Ryzen has 512MB. Other than that, neat. It is amusing that your CPU gets four times the cinebench score of my stock 3570K.
|
|
#
?
Mar 26, 2017 04:44
|
|
|
BurritoJustice posted:The CPU-Z benchmark is inaccurate for Ryzen, as it was made with 256MB L3 cache in mind and the Ryzen has 512MB. Other than that, neat. It is amusing that your CPU gets four times the cinebench score of my stock 3570K. Oh.  Well good to know that cpuz one is bogus.
|
|
#
?
Mar 26, 2017 04:50
|
|
|
BurritoJustice posted:The CPU-Z benchmark is inaccurate for Ryzen, as it was made with 256MB L3 cache in mind and the Ryzen has 512MB. Other than that, neat. It is amusing that your CPU gets four times the cinebench score of my stock 3570K. I really can't understand where people are coming from with this stuff, since actual workloads are not being designed around that.
|
|
#
?
Mar 26, 2017 05:00
|
|
|
BurritoJustice posted:The CPU-Z benchmark is inaccurate for Ryzen, as it was made with 256MB L3 cache in mind and the Ryzen has 512MB. Other than that, neat. It is amusing that your CPU gets four times the cinebench score of my stock 3570K. Cache sizes sure made a great leap forward since the morning, no?
|
|
#
?
Mar 26, 2017 05:37
|
|
|
Re-bench with faster rams and after the windows patch: quote:average of the averages Source is some reddit post so I have no idea how legit this is but if true, AMD seems to be back on the menu even for a gaming rig? I don't care much about the avg, but those 0.1 percentile numbers are very impressive. Makes me wonder what the poo poo watch doge is doing though 
|
|
#
?
Mar 26, 2017 09:07
|
|
|
Without knowing what graphics card was used a lot of these numbers may be gpu bottlenecks. I can't view the source video but the description on YouTube doesn't mention what graphics card is used.
|
|
#
?
Mar 26, 2017 09:39
|
|
|
Boiled Water posted:Without knowing what graphics card was used a lot of these numbers may be gpu bottlenecks. I can't view the source video but the description on YouTube doesn't mention what graphics card is used. He says it's a GTX 1070 clocked at 2200MHz with vram clocked at 8900MHz.
|
|
#
?
Mar 26, 2017 09:52
|
|
|
e:fb Agreed that the 1080 / Ti would have been more ideal.
|
|
#
?
Mar 26, 2017 09:56
|
|
|
SwissArmyDruid posted:e:fb Would have made the differences if any more pronounced. I guess the take home is save $20, buy a 7700k.
|
|
#
?
Mar 26, 2017 10:00
|
|
|
Boiled Water posted:Would have made the differences if any more pronounced. If you had pure gaming in mind.
|
|
#
?
Mar 26, 2017 10:08
|
|
|
wargames posted:If you had pure gaming in mind. Unless you specifically know that you are going to be doing something that spreads well across cores - Ryzen is not the answer. 6-cores worth of Haswell-E at 4.5 GHz, sure, that's viable in both gaming and productivity tasks. But in gaming - depending on your memory situation, Ryzen may perform at/below Sandy Bridge levels. Its single-core performance is pretty terrible. Again - remember that even Ryzen 1700 costs $320 per chip, which is 1700X pricing. I paid the same amount for a 5820K a year ago (before memory prices doubled) and I paid about a pretty decent chunk less than most Ryzen motherboards cost at launch ($140 minus $30 bundle discount). The 1700 is the only viable processor in the series at the current pricing. Ryzen 1800X? uh, not worth it, to say the least, way to lay out a half grand for an Engineering Sample-quality processor with a poo poo-tier memory controller/interconnect. It's most definitely not worth laying out nearly twice what a 7700K costs. Which is the super funny part of all these Ryzen launch reviews - they're comparing against a 7700K processor that costs half as much in the main, while focusing on how it kinda sorta keeps up with Intel's $1000 competitor in certain situations. Paul MaudDib fucked around with this message at 10:19 on Mar 26, 2017 |
|
#
?
Mar 26, 2017 10:10
|
|
|
wargames posted:If you had pure gaming in mind. We're talking about pure gaming benchmarks, it's at the forefront of my mind.
|
|
#
?
Mar 26, 2017 10:16
|
|
|
I think the Ryzen 5 6-cores with the high clocks out of the box (1600X, for example) are going to be much better value. When they refine the process more and get higher clocks out of these things, they're going to be good for every purpose. Single core performance of Ryzen is NOT terrible. Bulldozer's single core performance is terrible. HalloKitty fucked around with this message at 11:52 on Mar 26, 2017 |
|
#
?
Mar 26, 2017 11:46
|
|
|
HalloKitty posted:I think the Ryzen 5 6-cores with the high clocks out of the box (1600X, for example) are going to be much better value. When they refine the process more and get higher clocks out of these things, they're going to be good for every purpose. From what we know the r5 isn't a better process just r7 with cores disabled.
|
|
#
?
Mar 26, 2017 11:58
|
|
|
wargames posted:From what we know the r5 isn't a better process just r7 with cores disabled. I know, I'm simply suggesting that in future, if/when they do improve it to get the clocks up, it will be a formidable CPU.
|
|
#
?
Mar 26, 2017 12:04
|
|
|
I'm hopeful that zen+ or whatever terrible name they'll give it will be the actual gamechanger and not just a flotation device keeping team red afloat.
|
|
#
?
Mar 26, 2017 12:05
|
|
|
HalloKitty posted:I know, I'm simply suggesting that in future, if/when they do improve it to get the clocks up, it will be a formidable CPU. Ryzen 2.0 might be a lot better, ryzen 5 as it exists now is basically 4-core Bulldozer 2.0. golly when I think about ryzen's problems right now it's probably the Infinity Fabric, how about we make those plebs buy a bunch of 2x2 core chips instead of the 4x0 chips that might actually perform decently against - say - their i5 competitors instead of - say - the $50 bulldozers we'll try to offer instead? But I guess the fanbase will buy anything we pump out at the 4-core-for-$200 price point, right? i guess we can always fix it next stepping, right boys? Paul MaudDib fucked around with this message at 12:31 on Mar 26, 2017 |
|
#
?
Mar 26, 2017 12:25
|
|
|
Bulldozer 2.0 eh?
|
|
#
?
Mar 26, 2017 12:47
|
|
|
That does seem like hyperbole.
|
|
#
?
Mar 26, 2017 12:52
|
|
|
It's more like a painting of haswell where the architecture was described by talking through a very long cloth, an infinite fabric if you will.
|
|
#
?
Mar 26, 2017 12:55
|
|
|
Paul MaudDib posted:Ryzen 2.0 might be a lot better, ryzen 5 as it exists now is basically 4-core Bulldozer 2.0. Too real, Paul. Dial it back a tad, buddy. Too real.
|
|
#
?
Mar 26, 2017 13:03
|
|
|
B-Mac posted:Bulldozer 2.0 eh? hey bud - bearing in mind that the memory clock is intimately tied to this chip's perfromance - is there any hint yet this chip might be able to, say, reliably achieve the standard DDR4-3000 CL15 clocks that are actually available in commodity gaming memory? it's only been what, two months since launch? There is like one memory kit officially approved for this architecture now, right? Otherwise you get to slum it up at like DDR3-1600 speeds, right? And this also is a design feature that happens to cripple this architecture? if only there was an "interconnect fabric" that was bottlenecking us on these slowdowns, man that would be terrible (it'll probably be fixed in the next serious stepping/revision tho) wargames posted:That does seem like hyperbole. Do you actually care to offer a launch price for 2x2 Ryzen 5 then? 8-core Ryzen is $320, let's hear your launch price for 4-core. Significantly less than $200? For something that competes with a Pentium G4650 in gaming? Paul MaudDib fucked around with this message at 14:06 on Mar 26, 2017 |
|
#
?
Mar 26, 2017 13:53
|
|
|
I clearly need to do reading on infinity fabric and not carry over performance concepts from Intel chips. AMD performance really is tied to ram frequency in a stronger way than intel's "get to this line and anything faster is just diminishing returns" ?
|
|
#
?
Mar 26, 2017 14:12
|
|
|
I really don't see any real problem with 1700x for gaming at all and it has that benefit of being able to handle everything else that grown rear end people do on their computer as well. Seems like a fine offering to me. I do wonder how large is the audience of people who see higher min fps and lower frametimes at settings you usually play at but then see max fps being lower(but still perfectly fine) and decide that means something bad? I do understand benching at lowest settings/resolution to really see what you can squeeze out of it but does that even corelate to anything when you consider real life user experience for the average user(barring cases when a chip really bombs in said benchmarks ofc)?
|
|
#
?
Mar 26, 2017 14:26
|
|
|
TheCoach posted:I do understand benching at lowest settings/resolution to really see what you can squeeze out of it but does that even corelate to anything when you consider real life user experience for the average user(barring cases when a chip really bombs in said benchmarks ofc)? Turns out that no, not really. The idea of going low resolution and settings was, as far as I can tell, to emulate what happens when gpus get ever faster, supposedly so you know better how your poo poo will perform next year when you buy that beefy gpu that puts out so many frames. But some dude then found out that this isn't the case at all and the bulldozer shitshow of a chip that was 10% behind sandy bridge in games is now 10% ahead in real world scenarios. It's still behind at low resolutions and settings, though, so low resolution testing tells you how your junk will perform at low resolution in 5 years pretty accurately.
|
|
#
?
Mar 26, 2017 15:07
|
|
|
Paul MaudDib posted:Do you actually care to offer a launch price for 2x2 Ryzen 5 then? 8-core Ryzen is $320, let's hear your launch price for 4-core. Significantly less than $200? For something that competes with a Pentium G4650 in gaming? Since it isn't out yet i can't provide any benchmarks to disprove your claims 4 core zen is just another bulldozer.
|
|
#
?
Mar 26, 2017 15:25
|
|
|
Simulated 4 Core Ryzens, using the setup AMD is claiming will be for sale, are slightly behind the 6 and 8 core Ryzens CPU's and it consistently beats the Pentiums and i3s while sitting at a very competitive price. These benchmarks are in this thread, done by a reviewer Paul has trusted; he's just legit trolling or hyperventilating, I can't tell which, Poe's Law and all. 4 threads or even 4 cores are becoming a bottleneck nowadays and I seriously can't recommend an i3 or i5 due to Ryzen because of this. Pentium has a place because all signs point to AMD having no low end answer until RR and R3.
|
|
#
?
Mar 26, 2017 16:30
|
|
|
So, tell me which one of these is closer to true, because I'm not pulling a clear message here: A: Ryzen will sort out its hiccups in time, and will be a better choice than Intel for 16 thread use cases in price and maybe even performance. B: Ryzen will remain finicky and Intel will be a one-time "I don't feel like loving with this" tax away from equivalent performance, while dodging a dozen problems that could bring what you're doing to a screeching halt.
|
|
#
?
Mar 26, 2017 17:46
|
|
|
|
| # ? May 21, 2024 11:54 |
|
|
Wulfolme posted:So, tell me which one of these is closer to true, because I'm not pulling a clear message here: Nobody is. You'll just have to wait and see like the rest of us. Flip a coin if it makes you feel better.
|
|
#
?
Mar 26, 2017 17:52
|
|
