overclocking today's CPUs, specifically Ryzen

[CHADBOGA]

Ryzen is surely the worst overclocking CPU that either AMD or Intel have released for many a year.

If the problem is simply the process, then that may bode well for when they move to a different process.

 

[scannall]

Ryzen is surely the worst overclocking CPU that either AMD or Intel have released for many a year.

If the problem is simply the process, then that may bode well for when they move to a different process.

Overclocking is almost always process. LPP is an energy efficient process. And getting it to 4 Ghz is really quite good. The problem with processes, is that it's a giant money sink. At some point you're stuck with it for good or ill. 20nm comes to mind... Similar to architecture design, you will sink years of time and money, and not know if you've done great or shot yourself in the head until you have far too much sunk cost into it.

 

[Yeroon]

Ryzen is surely the worst overclocking CPU that either AMD or Intel have released for many a year.

If the problem is simply the process, then that may bode well for when they move to a different process.

Worst as in frequency, or worst as in percentage above stock?
Because it may not win frequencies (construction cores were great at that) its IPS is still fairly impressive even when its limited to around 4 Ghz due to process, and an OC from 3.1 base to 3.8 OC (say ryzen 7 1700) is a 22% ish gain, which is fairly impressive.

 

[VirtualLarry]

https://www.youtube.com/watch?v=cPWMIpVrVQ0

In this video, they overclock a Ryzen 5 1600, to 3.85Ghz @ 1.400V, and he has graphs near the end of the video showing max temp as 72C.

I've tried overclocking to 3.80Ghz @ 1.3500V, and I get 95C in Ryzen Master. And an eventual black-screen.

I re-pasted the heatsink, with MX-4, kind of a larger pea size, due to the larger heatsink. And the copper core had a hairline scrape in it, due to me scraping it with a screwdriver. It didn't appear to stick outwards though. I tightened the heatsink down as far as it would go this time.

Edit: Now that it has heated up, Ryzen Master is displaying 105C. I do have thermal shutdown enabled in UEFI.

 

[VirtualLarry]

I can't overclock this Ryzen 5 1600 rig on an ASRock AB350M Pro4 worth a hill 'o beans. I can use XMP 2400 settings, and that's about it.

I've tried 3.80Ghz at 1.3500V and 1.3750V, but temps are too high and I get black-screen.

TBH, I don't know why I'm getting black-screens, and not BSODs. Maybe it's really the video card (GTX 1050ti 4GB) that's failing on me? It doesn't spin the fan unless it's under load, and just sitting idle, in the same case with my Ryzen CPU, it may be overheating with the fan not spinning.

 

[CHADBOGA]

Overclocking is almost always process. LPP is an energy efficient process. And getting it to 4 Ghz is really quite good. The problem with processes, is that it's a giant money sink. At some point you're stuck with it for good or ill. 20nm comes to mind... Similar to architecture design, you will sink years of time and money, and not know if you've done great or shot yourself in the head until you have far too much sunk cost into it.

Will Threadripper's clock speed surprise for a 16 core CPU?

AMD's power consumption is quite impressive for Ryzen, but with that 4Ghz hard ceiling, whilst that hurts them on their lesser core models, they may not be particularly affected when the total number of cores is limiting clockspeed.

So I wonder if comparably, Threadripper will be their most competitive CPU against Intel alternatives?

 

[Ratman6161]

I can't overclock this Ryzen 5 1600 rig on an ASRock AB350M Pro4 worth a hill 'o beans. I can use XMP 2400 settings, and that's about it.

I've tried 3.80Ghz at 1.3500V and 1.3750V, but temps are too high and I get black-screen.

TBH, I don't know why I'm getting black-screens, and not BSODs. Maybe it's really the video card (GTX 1050ti 4GB) that's failing on me? It doesn't spin the fan unless it's under load, and just sitting idle, in the same case with my Ryzen CPU, it may be overheating with the fan not spinning.

Hmm. I have an Asrock AB350 Pro4 which is nearly the same as yours except full ATX and I have the same CPU that you do. For me temperatures are zero issues. At 3.8 Ghz and 1.35V I've never seen any temperature above 55c...but you say you are getting 95C? Something else is very wrong there. Since you have the AB350M can we assume this is in a small form factor case (mine is an Antec ThreeHundred mid tower case)? Is there adequate air flow through the case? If airflow through the case is poor and you've got a 1050TI in there, I can see the temps inside the case getting pretty high. One other observation - whenever i've been in a situation where my temperatures were much higher than everyone else with a similar rig, it turned out I didn't have the cooler mounted properly. And fans not spinning at all? I don't think I've seen a cpu or video card where the fan stops completely even at idle - sounds like a problem to me.

Another possible problem could be the power supply. Sometimes in small form factor cases (assuming thats what you've got due the the motherboard you are using) the power supplies tend to be wimpy. I looked up an MSI GTX1050Ti and MSI says it draws 75 watts (not bad really) and the minimum recommended power supply is 300 watts..but 300 wats may be pretty wimpy for what you are doing when you add in everything else in your system. I'm using a 600Watt PS. I've seen a lot of weird and difficult to diagnose problems crop up when a PS was either on its last legs or just to wimpy for the task.

 

[UsandThem]

And fans not spinning at all? I don't think I've seen a cpu or video card where the fan stops completely even at idle - sounds like a problem to me.

There are many different models of video cards where the fan stops when the GPU temperature is below a certain point. For example, my MSI card features "Zero Frozr" that stops the fans when the temp is below 60c. However, I manually set it to always run at 23% in MSI Afterburner to keep it below 50c (can't hear it at that RPM).

https://us.msi.com/Graphics-card/GeForce-GTX-1060-GAMING-X-6G.html#hero-overview

However, a CPU fan never stops (many motherboards will refuse to boot and immediately shut down if no fan is detected).

 

[Valantar]

There are many different models of video cards where the fan stops when the GPU temperature is below a certain point. For example, my MSI card features "Zero Frozr" that stops the fans when the temp is below 60c. However, I manually set it to always run at 23% in MSI Afterburner to keep it below 50c (can't hear it at that RPM).

https://us.msi.com/Graphics-card/GeForce-GTX-1060-GAMING-X-6G.html#hero-overview

However, a CPU fan never stops (many motherboards will refuse to boot and immediately shut down if no fan is detected).

A few (very few) newer CPU coolers have fans designed to stop entirely below a given PWM signal (usually 40%), like a few recent models from Arctic. Still, they're a tiny, tiny minority of coolers. I have no idea how a motherboard would react to that, though.

 

[UsandThem]

A few (very few) newer CPU coolers have fans designed to stop entirely below a given PWM signal (usually 40%), like a few recent models from Arctic. Still, they're a tiny, tiny minority of coolers. I have no idea how a motherboard would react to that, though.

True, there a few fanless coolers, but as you said not very many. The only way I would think it would work is to go into the BIOS and disable hardware monitoring (fan/temp). However, that's nothing I would ever do. If a person can hear their fan at 30-40%, they need a new cooler/fan. I can't hear mine until it above 800 - 900 RPM (depending on ambient noise level in room).

 

[VirtualLarry]

There are apparently a few threads (one recent), about "black screen" issues with Nvidia drivers and the Windows 10 Creator's Edition. But some of the other related threads, talk about a "black screen" at Windows boot time, with only a mouse cursor.

But I'm getting a black screen while crunching.

Anyways, I'm at 3700 Mhz and 1.2875V. Running 10 threads of BOINC, and Suspended GPU crunching, and it hasn't crashed yet, I've left it running since this morning and been away a few times.

So, now I'm suspecting a video card, or a thermal issue that the video card contributes to.

I was getting "PerfCap" of "Pwr" and "Vreg" (or "Vrel"?) at the same time. I don't know if that means that the GPU is limited in performance, because power limits are exceeding TDP, and the VRMs are becoming overloaded or overtemp, or what.

A YouTube review of my motherboard's VRMs by "Buildzoid", claimed that the VRMs should be easily capable of OCing an 8-core Ryzen 7 CPU, at 1.425V and 4.0Ghz. Which, perhaps, thermals permitting, might be possible.

But there's really NO airflow at the upper half of the case, which is likely to be my problem.

Edit: Not a SFF case, instead, a full ATX mid-tower case (a cheap AZZA with window from Newegg on ebay), and a ThermalTake TR2-430 (new version) ATX PSU, with five-year warranty.

 

[VirtualLarry]

Ok, this is weird. I downloaded HWMonitor, ran it, and it didn't give all the readouts. So I closed CPU-Z, GPU-Z, and Ryzen Master.

HWMonitor is giving VCORE readings of 2.624V, min 1.312V, max 2.624V.

WTF??

Is this the result of trying to use both BIOS and Ryzen Master to OC? CPU-Z gave my VCORE as I specified in BIOS, plus a tiny bit. (NOT 2+V)

Edit: Best I can tell, on my board, with the newest version of HWMonitor that supports Ryzen, it is reporting VCORE that is exactly double what it really is.

Which is strange, because CPU-Z displays it basically correctly.

And the UEFI agrees with what was set, in the monitoring screen, and CPU-Z, more or less.

I was a bit scared, after trying 1.41V, and not getting back into the BIOS after a few power-offs. (Needed to cool down?) I thought maybe, if HWMonitor was accurate, I had fried my Ryzen 5 1600, with 2.9V. Thankfully, that doesn't appear to be the case.

I am now running my 1600 stock, on the stock cooler, and I'm afraid that's where it's going to have to stay, until I get a liquid cooling setup.