Ryzen 7 2700 seriously slow

Page 11 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Topweasel

Diamond Member
Oct 19, 2000
5,436
1,654
136
how much ram is used? why not test all the ram if talking about stable overclock for ram. i guess i can just keep asking over and over how do you test memory that is in use by operating system already...

Memtest. But memtest is a sector by sector testing tool. But that tells you if you have bad memory. Not if it's not friendly with an overclock. When you are clocking DDR4 to 3000GHz. You are really clocking the memory bus to 1500MHz (Because DDR) you are not actually overclocking the memory chips themselves. There is a little more to it. When overclocking you aren't generally increasing the likelihood of increased bad sectors, but the strain on the memory subsystem.

So decide what you are doing
Bit tests. Memtest
Memory subsystem tests. One of a 100 different 24+ hour benchmark tools that do tons of memory calls.
 

B-Riz

Golden Member
Feb 15, 2011
1,482
612
136
how much ram is used? why not test all the ram if talking about stable overclock for ram. i guess i can just keep asking over and over how do you test memory that is in use by operating system already...

Low level whole module physical ram testing requires a special machine.

Memory in a modern OS is abstracted away, all you have to do is load up the majority of it to test the stability of over-clocks.

So, yeah, you may want to do some research before trolling us.
 

Wuzup101

Platinum Member
Feb 20, 2002
2,334
37
91
So I was poking around with my system today (have the same 2700 + X470 Taichi as the OP) to try to resolve another issue and I decided I'd see if my system behaves the same way as the OPs system behaves. I tried this using the latest two BIOS revisions for the Taichi (3.20 and 3.40, with 3.40 being the one that was just added for Ryzen 3 support at the end of May). Anyway, same behavior in both:

I'm currently running chrome, HWmonitor, CPUz, and 2 threads of FFT in Prime 95. Stock bios settings (basically auto everything) with the exception of: 1) using my known good XMP memory settings, fan profiles, and I have CSM disabled for booting UEFI only. All processor related voltages / timings are stock. HWmonitor is reporting my max clock across all 8 cores as 3499, min 2679, and right now through the prime 95 test they are all sitting at 3350 +/- a few mhz. In my observation this is the average. Processor utilization is <15%, and current package temperature (and also max package temperature) is 42 deg C (using a 2x120 AIO), max power draw on the package is 41W (but sitting more or less at 28W right now).

So it would seem that something here is NOT working as AMD intended or advertised. I'm going to go back to my manual overclock, but I just wanted to report that I do see basically what the OP is saying.

Edit: under 16 threads of small FFT (max heat) at 100% processor usage I'm at 3300-3325mhz, about 63W on the package, and <50C temps. This makes me realize how quiet a 3700X could be at full load haha.
 
Last edited:

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
HWmonitor is reporting my max clock across all 8 cores as 3499, min 2679, and right now through the prime 95 test they are all sitting at 3350 +/- a few mhz. In my observation this is the average. Processor utilization is <15%, and current package temperature (and also max package temperature) is 42 deg C (using a 2x120 AIO), max power draw on the package is 41W (but sitting more or less at 28W right now).

So it would seem that something here is NOT working as AMD intended or advertised. I'm going to go back to my manual overclock, but I just wanted to report that I do see basically what the OP is saying.

Edit: under 16 threads of small FFT (max heat) at 100% processor usage I'm at 3300-3325mhz, about 63W on the package, and <50C temps. This makes me realize how quiet a 3700X could be at full load haha.

But I think it is working as they advertised. As I posted this earlier in the thread, those speeds are in line when leaving the CPU at stock and relying on its pre-defined precision boost and XFR 2 states:

clock_analysis_compared.jpg
 

Wuzup101

Platinum Member
Feb 20, 2002
2,334
37
91
But I think it is working as they advertised. As I posted this earlier in the thread, those speeds are in line when leaving the CPU at stock and relying on its pre-defined precision boost and XFR 2 states:

clock_analysis_compared.jpg

Yeah I'm honestly not sure if it is or isn't. Frankly, it doesn't really impact me, because I am manually overclocking anyway. However, the behavior that I'm seeing just basically looks like a flat line across the page. I was fully aware of that chart before I purchased, which was one of the reasons I saved the $70 and went with a non X sku (and it honestly was just a placeholder for a 3rd gen Ryzen anyway.

Their article also says that when idle the base clocks down to 1550mhz... I'm not seeing that... it was pretty much always at 3200mhz - 3500mhz with very few exceptions.

https://www.techpowerup.com/reviews/AMD/Ryzen_7_2700/16.html

I'm also not seeing clocks that are equal to theirs with 16 cores fully loaded. Infact, I'm seeing less running 1 thread of prime than they did with 16 cores fully loaded (3420mhz vs my 3350mhz), and I'm well under the TDP for the processor and have way better cooling than the stock HSF (though I'm not sure what cooling solution they are using). I did try various loads, but wasn't able to replicate anything nearing a 4.1ghz single core turbo.
 

UsandThem

Elite Member
May 4, 2000
16,068
7,380
146
Yeah I'm honestly not sure if it is or isn't. Frankly, it doesn't really impact me, because I am manually overclocking anyway. However, the behavior that I'm seeing just basically looks like a flat line across the page. I was fully aware of that chart before I purchased, which was one of the reasons I saved the $70 and went with a non X sku (and it honestly was just a placeholder for a 3rd gen Ryzen anyway.

Their article also says that when idle the base clocks down to 1550mhz... I'm not seeing that... it was pretty much always at 3200mhz - 3500mhz with very few exceptions.

https://www.techpowerup.com/reviews/AMD/Ryzen_7_2700/16.html

I'm also not seeing clocks that are equal to theirs with 16 cores fully loaded. Infact, I'm seeing less running 1 thread of prime than they did with 16 cores fully loaded (3420mhz vs my 3350mhz), and I'm well under the TDP for the processor and have way better cooling than the stock HSF (though I'm not sure what cooling solution they are using). I did try various loads, but wasn't able to replicate anything nearing a 4.1ghz single core turbo.

It could all just come down to them using a different motherboard (they used a MSI X470 Gaming M7 AC). I know my Z270 Taichi displays odd behavior with a 7700K (I just stayed with the 6700K instead), not hitting its advertised speeds. So the difference could be as simple as something like that.

But it really sounds like the OP wanted 2700X performance, but didn't want to buy one (or overclock his 2700), and now has buyer's remorse on their purchase decision. They just seem to make the Ryzen 2700 CPU sound like its performance is Cyrix-like when it seems they maybe didn't need to upgrade from an overclocked 2500K (which is still a decent CPU all these years later). They said they couldn't wait for the Ryzen 3000 series, and that seems to be where they would have seen a big difference coming from that Sandy Bridge CPU.
 
  • Like
Reactions: Wuzup101 and ZGR

Wuzup101

Platinum Member
Feb 20, 2002
2,334
37
91
This is a 65w CPU and it is very good at staying at 65w with stock settings.

This review also experiences the same low clocks before manually overclocking:

https://www.hardocp.com/article/2018/06/01/amd_ryzen_7_2700_overclocking_review/5


This is a bad comparison, but my CPU is also a 65w CPU. If I don't increase the power limits, it throttles like crazy when overclocking.

So that's interesting though, they said they saw 3.8 to 4.1 ghz under a single core cinebench load. I actually just ran Cinbench R20 on both my desktop and laptop for ***** and giggles. This is using the auto clock settings, not the desktop OC settings in my sig:

Ryzen 2700 (default bios, ram running 3200mhz 14-14-14-31, custom fan profiles, UEFI boot changes only):
CPU: 3207
CPU (Single Core): 335

Intel 8850h (2.6 base, 4.3 turbo stock, 6c/12t 45w tdp; -0.120 vcore w/ 4.4ghz set as all core turbo in intel XTU)
CPU: 2539
CPU (Single Core): 455

What did I see when I watched these?
- The 2700 was pegged at 3325mhz during the all core test staying below 65w tdp and 3350 during single core. It didn't boost higher than that during single core. It also ran at 35C during the single core test and used nowhere near its 65w TDP. An no point does this processor seem to want to boost to similar levels that I've seen in reviews. It also doesn't want to downclock. I'm at a reported 0-1% processor usage while typing this, cores all sitting around 3350mhz, sometimes you get one that drops to 2700ish mhz.

- The intel processor maintained 4.4 for nearly the whole single core test, but was in the low to mid 3xxx range for the all core test. It was thermal throttling a little during single core (it's in a fairly thin laptop) but held 4.4 pretty well. It power limit throttled pretty hard during the all core test, which is to be expected. Unfortunately I can't get power limit tweaks to stick in XTU with the Lenovo X1E.

- In the article that you referenced, the 2700X only beat the 2700 by 7% on the Cinebench R15 single core test. Where as, the comparable systems that Cinebench R20 shows me, a 1700X beats me by 12%. Single core, I'm actually closer to a 2.3ghz i7-4850hq (298) than I am to a 1700X (378) at 335.

It could all just come down to them using a different motherboard (they used a MSI X470 Gaming M7 AC). I know my Z270 Taichi displays odd behavior with a 7700K (I just stayed with the 6700K instead), not hitting its advertised speeds. So the difference could be as simple as something like that.

But it really sounds like the OP wanted 2700X performance, but didn't want to buy one (or overclock his 2700), and now has buyer's remorse on their purchase decision. They just seem to make the Ryzen 2700 CPU sound like its performance is Cyrix-like when it seems they maybe didn't need to upgrade from an overclocked 2500K (which is still a decent CPU all these years later). They said they couldn't wait for the Ryzen 3000 series, and that seems to be where they would have seen a big difference coming from that Sandy Bridge CPU.

I don't necessarily disagree; however, I can say that I definitely do NOT believe this chip/board combo is performing correctly. I should also note, I'm just continuing screwing around with this because I think it's odd behavior. Anyone have any tests that they want me to run?

I guess, bottom line, I would have expected a relatively popular, mature board to not have these issues. This is the latest BIOS revision of a board that came out, what, 14 months ago?
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
I guess then, I owe an apology to the OP for being so harsh with my wording. Seems like there's some sort of incompatibility or poor CPU support going on here, between ASRock, X470 TaiChi, and the AMD Ryzen R7 2700 CPU.

I have had no real issues with mine, but as I mentioned, I'm using an Asus B450-F board.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
I guess, bottom line, I would have expected a relatively popular, mature board to not have these issues. This is the latest BIOS revision of a board that came out, what, 14 months ago?

Your CPU is still running ~125 MHz faster than the OP's in all-core turbo. That being said, ASRock seems to have terrible legacy UEFI support for their boards, at least on X370/X470, in various different ways. My X370 lost RAM OC headroom with newer AGESA versions. Not sure if the X470 Taichi has similar problems with the 2700x for which it was intended. Seems like Asus boards (like the Crosshair VII Hero) have better long-term UEFI support.