Question Ryzen 7 1700x overclocking

Feb 8, 2019
58
1
35
0
Just got a ryzen 7 1700x and a gigabyte X470 aorus gaming 7, 16gb Corsair lpx ddr4 Ram and an h100i. How would I go about overclocking this cpu to as high as it can safely go (3.8-4.0ghz) to get the most performance out of my rtx 2070?
 
Boot into your bios and set your CPU multiplier to 38 (3.8ghz) and set the v-core to 1.2. That will run your chip at 3.8ghz with a CPU voltage at 1.2. Boot the system, if it boots, then run a Cinebench benchmark. If your system crashes, then boot back into your bios and up your voltage to 1.21. Rinse and repeat by increasing the voltage by .01 till it can run the benchmark.

Watch your temps through Ryzen Master when you run the bench. For a daily driver, I like my average load temps to stay below 80C. So if the average temps go above 80C during the bench, then I would lower the voltage or the clockspeed.

If you want 4ghz, then just change the mulitplier to 40. I suspect your chip will run at 4ghz on all cores with your cooler just fine.
 
Feb 8, 2019
58
1
35
0
Boot into your bios and set your CPU multiplier to 38 (3.8ghz) and set the v-core to 1.2. That will run your chip at 3.8ghz with a CPU voltage at 1.2. Boot the system, if it boots, then run a Cinebench benchmark. If your system crashes, then boot back into your bios and up your voltage to 1.21. Rinse and repeat by increasing the voltage by .01 till it can run the benchmark.

Watch your temps through Ryzen Master when you run the bench. For a daily driver, I like my average load temps to stay below 80C. So if the average temps go above 80C during the bench, then I would lower the voltage or the clockspeed.

If you want 4ghz, then just change the mulitplier to 40. I suspect your chip will run at 4ghz on all cores with your cooler just fine.
Any advice on voltage for 4.0ghz ? Can you share any of your experiences ? I’m coming from an fx 8350 that I oced to 5ghz so I’m very familiar with that type of overclocking is it similar ?
 
Feb 8, 2019
58
1
35
0
It is quite rare for a Ryzen 7 to hit 3.8ghz at 1.2v, more like 1.3ish volts. But it is still worth a shot, you might have a good chip there.

I lost the silicon lottery lol, I need 1.365 to push 3.8ghz.
What about 4.0ghz ? 1.365 doesn’t seem like too much my fx 8350 was on 1.45v at 5ghz
 
What about 4.0ghz ? 1.365 doesn’t seem like too much my fx 8350 was on 1.45v at 5ghz
I would keep the voltage below 1.4. I am actually fairly conservative and would stay below 1.35, but that is just me. In the end, high temps will kill the CPU quicker and higher volts correlate to higher temps. As long as your temps are okay, you CPU will be okay.

95C is when the CPU starts to throttle and that is very high. I would not keep it in the 90's for the long term or you will damage your chip over time. For a daily driver (daily use CPU), I would keep the load temps below 80C.
 
At 95c it should shut down, even 80c is not good. AMD's recommendations are 1.425v and 75c maximum.
The crazy thing is knowing what the real core voltage is. The only way to be sure is to measure it at the base of the CPU socket with a DVM. And don't forget proper measurement technique: use a ground reference as electrical close as possible to the measurement point.

Anybody who's going to be pushing AMD recommendations on their motherboard should make that measurement first. Using VCore alone is pretty much a 'fool-proof' way to be safe since it can be up to 140mV (or at least that's what I was told is the SVI2 spec for maximum motherboard/socket voltage sag) more than the actual core voltage. How much more depends on how/where a specific motherboard reads out VCore and reports it to utilities.

Also: I thought the CPU would start throttling at 95C...or is that throttling at 90 and shutdown at 95?
 
Yes, HWinfo has best data but most important thing is to look at average voltage under normal but prolonged load as higher voltages can be pretty transient and happen only for short time not influencing temps much and less harmful.
 
Yeah real core voltage is a nightmare to find on some monitors, so use HWINFO64 and use the SVI2 TFN sensor which shows exact vcore which is great.
That's what I've been using, but i only trusted it after I made the DVM measurements at the socket for my board.

But I did need that confirmation first. When the CPU's heavily loaded, as in a stress test, the SVI2 TFN on my 3.95Gig 1700 is reading very rational. A VCore reading 1.472 solid line, should be scary, but SVI2 reading 1.36-1.38 says it's it's a safe level at the CPU. But with no load on the CPU SVI2 goes up to 1.419-1.425...except VCore is reading in 300-500mV!

I had to go check it out, at the base of the socket with a DVM, where sure enough it's running around 300-500mV idle. So even if the CPU thinks it's getting 1.425? no it's not really; it's only getting a fraction of that at idle. That's all the VRM controller is sending out.

SVI2 TFN isn't a measured voltage, it's an interpolated voltage the CPU thinks it's getting. And it seems that, like the interpolated temperatures it sends out, it's wildly innaccurate when the processor is at idle. That's all i can figure it to be. Why? not sure but that seems to be what happens.
 
@drea.drechsler are you sure you're looking at Vcore, not VID or something?
I'm positive it's VCore: it's a reading in the HWInfo section for the Nuvoton monitoring chip. It also corresponds quite accurately with DVM measurements I take at the output of the VRM (the inductor leads, very easy to identify) and at all operating points: idle, full load and fluctuating loads. Fluctuating loads are hard to make comparison, though, as there is inevitable lag in the HWInfo readouts.

VID is displayed, in HWInfo, in the section for 'CPU [#0]: AMD Ryzen....' and is shown for each core discretely.

Easily done to keep those separate.
 
Reactions: TJ Hooker

ASK THE COMMUNITY

TRENDING THREADS