AMD Ryzen Discussion Thread.

[the54thvoid]

I would like to know how many people on TPU have managed a stable 4GHz OC on a 1700, and what MB, RAM, and cooler you used, and how long it took to get there? Also did you use auto or manual CPU voltage, and if manual, what voltage did you settle on?

There's a few review sites making it sound easy even with CPU voltage set to auto, but I'd rather hear it from end users.

Hell, I'm not happy pushing to 3.9Ghz on my 1700X. Takes a lot of volts and as good as my air cooler is, it's hard to be confident.

However, the thing boosts all cores to 3.8 (with 3.9) on two cores under heavy loading on stock settings. 1002 BIOS.

As for @Johan45 , they're a OC guru, i don't have the skills or patience for what they do.

 

[Johan45]

What MB, and are you running the latest BIOS version?

Like I said it was the CHVI ver 1002. I feel this one might need slightly higher voltage but it works fine. No real issues with mem etc.. It is my second, the first one lasted about a day and a half before the BIOS bug got it. That's not an issue any more. If I were to run this chip (lesser of 2 1700's) 24/7 it would be around 3.8, the other does 3.9 with ~ the same voltage. The Ryzen voltage seems to fluctuate a lot in CPUz etc.. I have found HWinfo64 seems to be the most accurate ATM the author has been working hard to keep it up to date. This pic is 1.35v BIOS set and I have marked the "actual" voltage in monitor for reference. The p95 test was ~ 80 minutes

This ones for you 54thvoid

 

[Caring1]

This ones for you 54thvoidView attachment 85940

 

[Frag_Maniac]

Like I said it was the CHVI ver 1002. I feel this one might need slightly higher voltage but it works fine. No real issues with mem etc.. It is my second, the first one lasted about a day and a half before the BIOS bug got it. That's not an issue any more.

Still though, not at all the results I'd be shooting for, so that makes me hesitant. I'd be looking for a solid 4GHz at 3200 RAM speed, and with 3200 rated RAM, not 3600. I might be OK with CAS 16 RAM, but prices will no doubt drop on the B-dies by the time I'm ready to buy, which may not be until end of year.

I just get the feeling Ryzen needs an expensive MB and RAM combo to OC a 1700 to 4GHz at 3200, and that's not at all appealing. It seems the CCX interconnect needs a lot of work. It certainly wouldn't be the first time AMD's first gen of a CPU has significant flaws.

Another reason I'm hesitant is I already got burnt on my i7-950 not OCing worth a damn, and going from 3 to 4GHz is quite a jump,. It's just that a few sites have made it sound like it's easy to do with a 1700, even with auto voltage setting.

On the other hand I can't see paying $170 more just to get a 600MHz higher clock.

 

[the54thvoid]

Like I said it was the CHVI ver 1002. I feel this one might need slightly higher voltage but it works fine. No real issues with mem etc.. It is my second, the first one lasted about a day and a half before the BIOS bug got it. That's not an issue any more. If I were to run this chip (lesser of 2 1700's) 24/7 it would be around 3.8, the other does 3.9 with ~ the same voltage. The Ryzen voltage seems to fluctuate a lot in CPUz etc.. I have found HWinfo64 seems to be the most accurate ATM the author has been working hard to keep it up to date. This pic is 1.35v BIOS set and I have marked the "actual" voltage in monitor for reference. The p95 test was ~ 80 minutes

View attachment 85939

This ones for you 54thvoidView attachment 85940

Like I say. Bloody Guru!

Edit: 1.4v won't get me along at 3.9Ghz, temps rise to 75 ish. SoC at 1.2 and LLC at 2. Prime crashed after a few mins.
Running same settings now but reduced clock to 3.8Ghz. Temps at 71. Noticed air coolers heat fast but tend to cool a few degrees. Water always crept up until the temps stabilised.

 

[Lt_JWS]

A surprise came for me today....

Ryzen 7 1700 vs Ryzen 5 1400 HSF

A quick OC 3.9Ghz @ 1.35vcore 2400Mhz ram

 

[RealNeil]

I was gonna start my Ryzen 1700X build tonight but Newegg sent me 2400MHz RAM instead of 3200MHz.
I have some 3200MHz. Geil RAM in my i5-6600K system so maybe I'll trade them out for a while.

Has anyone used GEIL RGB 3200MHz RAM in a Ryzen build yet? (or am I breaking new ground?)

Also, I ~could~ strip 16GB of GSKill 3200MHz RAM out of my 6700K build if I had to, but I'd rather leave that one alone if I can.

 

[HTC]

I was gonna start my Ryzen 1700X build tonight but Newegg sent me 2400MHz RAM instead of 3200MHz.
I have some 3200MHz. Geil RAM in my i5-6600K system so maybe I'll trade them out for a while.

Has anyone used GEIL RGB 3200MHz RAM in a Ryzen build yet? (or am I breaking new ground?)

Also, I ~could~ strip 16GB of GSKill 3200MHz RAM out of my 6700K build if I had to, but I'd rather leave that one alone if I can.

Check your memory here: http://rymem.vraith.com/

 

[RealNeil]

Thanks for the link HTC.

My GEIL 3200MHz isn't listed yet for the Crosshair board but I'll try it out just to see how it does.
The GSKill 3200MHz RAM I have ~is~ listed, so I know I have something that will work.

 

[HTC]

Thanks for the link HTC.

My GEIL 3200MHz isn't listed yet for the Crosshair board but I'll try it out just to see how it does.
The GSKill 3200MHz RAM I have ~is~ listed, so I know I have something that will work.

No prob.

That doesn't mean it wont work: just that it isn't tested. That's what i think it means.

You can also check if it works on other AM4 boards instead of your current one.

 

[Lt_JWS]

Loving the R5 1400 so far, it doesn't overclocked as good but it's an amazing value. Currently running Realbench @ 3.8ghz 1.375vcore. I will say this, it's running as hot as the R7 1700, I may just have bad contact on my H60 Gaming so far is nearly identical to the R7.
Got it from here,
://www.provantage.com/service/searchsvcs?QUERY=Ryzen&SUBMIT.x=0&SUBMIT.y=0

 

[RealNeil]

I already pulled the GEIL RAM out of the EVGA board.
I'll try it out first.
If it works well I'll probably get two more 8GB sticks of it for the Ryzen build.
I keep reading that Ryzen likes faster RAM.

 

[HTC]

I already pulled the GEIL RAM out of the EVGA board.
I'll try it out first.
If it works well I'll probably get two more 8GB sticks of it for the Ryzen build.
I keep reading that Ryzen likes faster RAM.

Be carefull: 4 sticks are allot trickier to get working then 2 and they tend to work @ lower speeds too, when it comes to Ryzen. @ least for now anyway.

That doesn't mean it wont work, but it is a bit harder, atm.

Also: dual rank RAM is much trickier to get working on Ryzen but it CAN be done (pic is hard to see).

 

[Hotobu]

So are there no plans to support ram speeds > 2933, or are MB manufacturers just not there yet?

 

[HTC]

So are there no plans to support ram speeds > 2933, or are MB manufacturers just not there yet?

From what i can tell, the problem is with the AGESA version that the boards have. Notice how even with the best high end boards you can't adjust memory subtimings. BIOSes need to mature: what doesn't work right now may work next week ...

AMD really screwed up in this department: their desire to have things tight-lipped led to this awful situation and the board makers should have the chips available a heck of allot sooner to prepare for a propper launch. It ended up hurting AMD allot, IMO.

 

[RealNeil]

Be carefull: 4 sticks are allot trickier to get working then 2 and they tend to work @ lower speeds too, when it comes to Ryzen. @ least for now anyway.

That doesn't mean it wont work, but it is a bit harder, atm. Also: dual rank RAM is much trickier to get working on Ryzen but it CAN be done (pic is hard to see).

It will all be trial and error for me with this build. I guess that if I want more than 16GB of RAM I may have to buy two 16GB sticks?
Whatever. I'll start the build tomorrow and we'll see how it goes.

 

[phanbuey]

wrong thread but LOL.

 

[Kanan]

It ended up hurting AMD allot, IMO.

Well it's not really that important on the long run, because this architecture will run for many years. CPU's are a different matter, compared to GPUs for example, where such a "problem" can hurt the sales inevitably. I think Ryzen will be a big success on the long run (especially compared to FX).

 

[HTC]

Well it's not really that important on the long run, because this architecture will run for many years. CPU's are a different matter, compared to GPUs for example, where such a "problem" can hurt the sales inevitably. I think Ryzen will be a big success on the long run (especially compared to FX).

When you're trying to make a comeback in a big way, any unnecessary problems are to be avoided and this particular memory issue is both unnecessary and avoidable, if only they just provided the mobo manufacturers with the required material for them to remove most of the kinks by launch day.

Instead, most RAM doesn't work @ advertised speeds, if @ all, and BIOSes are a general mess, mostly: as far as memory is concerned, that is.

 

[Kanan]

When you're trying to make a comeback in a big way, any unnecessary problems are to be avoided and this particular memory issue is both unnecessary and avoidable, if only they just provided the mobo manufacturers with the required material for them to remove most of the kinks by launch day.

Well it's easy for you to say. AMD is a way smaller company than Intel, people seem to forget that all the time. A CPU launch is a big thing and AMD is not big atm. The way I see it, they were under a hell lot of a pressure to release Ryzen, that's why the release didn't go "perfect" (what in life is perfect anyway?). That all said, Intel didn't had perfect launches too, people seem to forget that as well. But when did Intel exactly launch a new architecture? Ah yeah, almost TEN years ago! Yes, the Core architecture is that old.

"Bioses are a general mess." Okay, then again I'd bet, if I'd lay my hands on it, everything would work fine. I'd simply buy the right MB, the right Ram and do the right things. Why? Because I'm not an early buyer. I simply watched and learned a fuc* lot by now about Ryzen, that's why I'm so sure I can avoid the mistakes others did that simply didn't knew better. My edge always is and always was knowledge. But whatever, I digress. In the end, all I wanted to say is this: "problems" of Ryzen are relative and I don't see them as a (big) problem, while others do.

 

[HTC]

Well it's easy for you to say. AMD is a way smaller company than Intel, people seem to forget that all the time. A CPU launch is a big thing and AMD is not big atm. The way I see it, they were under a hell lot of a pressure to release Ryzen, that's why the release didn't go "perfect" (what in life is perfect anyway?). That all said, Intel didn't had perfect launches too, people seem to forget that as well. But when did Intel exactly launch a new architecture? A yeah, almost TEN years ago! Yes, the Core architecture is that old.

"Bioses are a general mess." Okay, then again I'd bet, if I'd lay my hands on it, everything would work fine. I'd simply buy the right MB, the right Ram and do the right things. Why? Because I'm not an early buyer. I simply watched and learned a fuc* lot by now about Ryzen, that's why I'm so sure I can avoid the mistakes others did that simply didn't knew better. My edge always is and always was knowledge. But whatever, I digress. In the end, all I wanted to say is this: "problems" of Ryzen are relative and I don't see them as a (big) problem, while others do.

There was an unnamed manufacturer that complained about the lack of time between getting the samples to start working on BIOSes options and launch day. Why didn't AMD provide mobo manufacturers with samples @ an earlier date? Was the need for secrecy that high?

Giving the mobo makers more time would help them have better RAM compatibility from the start, IMO. That doesn't mean they would be problem free but i think having more time to work on something before launching it worldwide would be beneficial because it tends to minimize problems.

 

[Kanan]

There was an unnamed manufacturer that complained about the lack of time between getting the samples to start working on BIOSes options and launch day. Why didn't AMD provide mobo manufacturers with samples @ an earlier date? Was the need for secrecy that high?

Giving the mobo makers more time would help them have better RAM compatibility from the start, IMO. That doesn't mean they would be problem free but i think having more time to work on something before launching it worldwide would be beneficial because it tends to minimize problems.

Yeah I'm not really arguing this and I know it anyway, I just ask people to have more understanding for the problems that AMD had and still have.

 

[the54thvoid]

My system worked fine from start. No hardware issues. Overclocking is not as generous as Intel but that doesn't make it bad.
W10 is way worse than Ryzen.

 

[hat]

There was an unnamed manufacturer that complained about the lack of time between getting the samples to start working on BIOSes options and launch day. Why didn't AMD provide mobo manufacturers with samples @ an earlier date? Was the need for secrecy that high?

Giving the mobo makers more time would help them have better RAM compatibility from the start, IMO. That doesn't mean they would be problem free but i think having more time to work on something before launching it worldwide would be beneficial because it tends to minimize problems.

I think secrecy was thrown out the window a long time ago. What about all those PR articles and "leaks"? Everything AMD said about Ryzen (PR) lined up with the "leaks" which also lined up with real reviews. That doesn't sound very secret to me.

Yeah I'm not really arguing this and I know it anyway, I just ask people to have more understanding for the problems that AMD had and still have.

Well it's easy for you to say. AMD is a way smaller company than Intel, people seem to forget that all the time. A CPU launch is a big thing and AMD is not big atm. The way I see it, they were under a hell lot of a pressure to release Ryzen, that's why the release didn't go "perfect" (what in life is perfect anyway?). That all said, Intel didn't had perfect launches too, people seem to forget that as well. But when did Intel exactly launch a new architecture? Ah yeah, almost TEN years ago! Yes, the Core architecture is that old.

"Bioses are a general mess." Okay, then again I'd bet, if I'd lay my hands on it, everything would work fine. I'd simply buy the right MB, the right Ram and do the right things. Why? Because I'm not an early buyer. I simply watched and learned a fuc* lot by now about Ryzen, that's why I'm so sure I can avoid the mistakes others did that simply didn't knew better. My edge always is and always was knowledge. But whatever, I digress. In the end, all I wanted to say is this: "problems" of Ryzen are relative and I don't see them as a (big) problem, while others do.

I agree with most of this. While part of me thinks AMD could possibly have done a better job of making sure their product was more polished at launch, avoiding most of the current issues we're looking at, another part of me knows that they really needed a successful product after all these not-so-successful years. A lot of people were excited for Ryzen, but people only can wait for so long before they say "fuck it, I'm just going with Intel, I'm done waiting". I'm also sure they wanted to get it out the door ASAP so they could start making money on it sooner rather than later. And yet another part of me has been hanging around here long enough to know better than to jump right into a brand new product. Early adopter issues are not an uncommon thing.

AMD finally has a good product that can properly compete with Intel now. Once those early adopter issues get worked out, it'll get even better. I'm happy for AMD and I hope they get a decent piece of the pie that Intel's been hogging all these years, and I hope to have a Ryzen system myself in the not too distant future if at all possible.

My system worked fine from start. No hardware issues. Overclocking is not as generous as Intel but that doesn't make it bad.
W10 is way worse than Ryzen.

W10 is way worse than a lot of things.

 

[bencrutz]

When you're trying to make a comeback in a big way, any unnecessary problems are to be avoided and this particular memory issue is both unnecessary and avoidable, if only they just provided the mobo manufacturers with the required material for them to remove most of the kinks by launch day.

Instead, most RAM doesn't work @ advertised speeds, if @ all, and BIOSes are a general mess, mostly: as far as memory is concerned, that is.

actually, most RAM works OOTB at speed that AMD supported officially on ryzen