Question New i5-13600k Build Cinebench R23 Score concerns

Nov 30, 2022
6
0
10
So I just built a new system:

CPU: i5-13600k
Cooler: Deepcool AK620
Motherboard: Asrock Z690 Steel Legend
RAM: 32 GB Silicon Power 3200 Mhz CL-16 DDR4
PSU: MSI MPG A1000G
Case: Antec NX260
GPU: Dell RX 580 8GB (brought from my old build can't upgrade the GPU yet)
4x 120mm Case fans

I just put this PC together and haven't messed with many settings aside from PL2. When I run Cinebench R23 for single core I get 1970-2002, which lines up with what I have seen reviewers post, but when I run the multicore test the highest result I have gotten is 20735, which is more thank 2k under the UNDERCLOCKED score that is shown here:
View: https://www.youtube.com/watch?v=h7ZlCM-QLU4&t=317s
I use Open Hardware Monitor to monitor my temps and during Pass 1 or Pass 2 my temps spike to around 95-100c for about a second or two while the fans spin up and then my temps stabilize at 65-68c for the entire rest of the test, so I don't think that my cooler is seated incorrectly. I'm concerned something is wrong with my cpu if its near stock settings have a benchmark almost 10% worse than a underclocked cpu in the same sku. Any ideas?
 
So you haven't enabled XMP for your memory yet in the BIOS? That's likely where you're missing performance. Also make sure you have the latest BIOS version for your motherboard.

Also, if you haven't already, it would be advisable to go the product support page for your motherboard and download/install the latest chipset (.inf), network adapter (LAN and/or WiFi) and audio controller drivers. Don't rely on Windows native drivers, ever, unless there is absolutely no other choice.
 
Nov 30, 2022
6
0
10
So you haven't enabled XMP for your memory yet in the BIOS? That's likely where you're missing performance. Also make sure you have the latest BIOS version for your motherboard.
Running latest version of BIOS. XMP 2.0 is set to 3200 which is what my kit is rated for and I have it set to Gear 1. I have adjusted power limits a few times, PL1 is set to 125w which is what it came out of the box at and I have PL2 at 200W, I have tried between 181W and 225W and haven’t seen a difference in scores, just a slight difference in temps during the initial spike. For thermal management I set the fans on my AK620 to full speed for the duration of one of the tests and it didn’t affect the results either. No other BIOS settings have been changed from default.
 
Ok. Great. Did you install the latest motherboard drivers. These:

Realtek audio driver: https://download.asrock.com/Drivers/All/Audio/Realtek_Audio(v9231.1_UAD_WHQL_Nahimic).zip

Intel Z690 chipset driver: https://download.asrock.com/Drivers/Intel/INF/INF(v10.1.19199.8340).zip

Intel management engine driver (Pretty much a requirement for proper operation on Intel systems these days): https://download.asrock.com/Drivers/Intel/Others/ME(v16.1.25.1917_CONS).zip

Realtek LAN driver: https://download.asrock.com/Drivers/All/LAN/Realtek_LAN(v10053_10062021).zip

And do you have the ASRock motherboard utility or any other "ASRock utilities" installed? If you do, uninstall them and never install them again. ALL of the motherboard manufacturer utilities like Armory crate, MSI dragon center, etc. should be avoided except where it is absolutely necessary for RGB control and even then you should make sure there isn't a satisfactory open source alternative available, which usually there is, and they usually work a lot better than the OE solution anyhow. All these utilities tend to cause far more problems than they solve.

And, was this a CLEAN install of Windows, specifically for this new hardware? You aren't using a continuation of any other Windows installation that was previously used with another system, right?
 
Nov 30, 2022
6
0
10
All drivers up to date, clean install of windows. The only Asrock software is the Auto Driver installer that the motherboard installed on the first boot up to install updated drivers, I did not download the “Asrock Utility” on the driver page. This time it got a score of ~20900 so within margin of error of the scores I have been getting I think.
 
Nov 30, 2022
6
0
10
I should note that I am doing the “10 Minute (Throttling)” version of the multi-core test, and I was unaware until today that several of the scores reviewers have posted of 23-25k have apparently been with “Off” selected as the minimum duration of the test. I have no experience with benchmarking, so I am not sure if that matters or not in this case.
 
That's the thing about benchmarking. What one person's system gets, even with identical hardware, might be COMPLETELY different from what the next person gets depending on how the memory timings and BIOS settings are configured, how clean the OS installation is, variations in sample quality between CPUs, memory and even motherboards, drive speed and performance (Which too is affected by how empty or full the drive is on top of it's paper specs). Just lots of things that don't make for a direct apples to apples comparison and THAT is assuming all hardware is identical models.

The minute you add different hardware into the mix, you really can't directly compare your performance to anything other than yourself. Get a baseline, start making changes and tweaks or trying to apply resolutions to certain problems and compare the results to the baseline as you go.

As well, you should download and install HWinfo, run "sensors only" and leave it running the next time you run your multicore and single core benchmarks. Afterwards, take screenshots of the HWinfo sensors which will likely require taking three or four screenshots while scrolling down in between shots to get them all. This might show us something in the way of clues if something is off.
 
Nov 30, 2022
6
0
10
How do I post screenshots to this forum? It looks like the only options for posting screenshots are linking to other websites. On a side note I went into Cinebench and toggled on the "Advanced" mode that let me select "Off" for the minimum duration and immediately got a 23.5k score. It seems the reviewers I have seen actually show performing the test use this mode and got a comparable score within a couple hundred points, so I may be able to chock this one up to inexperience with benchmarking. I downloaded the HWinfo64 you recommended and in both the single pass test and a 10 minute throttle test, the only "Yes" that it showed related to power limits being exceeded or being the limiting factor. Specifically "Core Power Limit Exceeded" said "Yes" on all cores and "Package/Ring Power Limit Exceeded" said "Yes." Max temp was 101c but average temp was 67c. It looks like the only reason my 10 minute throttle test only gets a score of 20.5-20.9k is due to the power limits I have set for PL2 because it hits 225W which is my PL2 limit and then drops back down. I ran the multicore single pass test multiple times and got ~23.5k+ each time so it seems everything is working correctly after all, I just don't know a thing about benchmarking or advanced hardware usage. I'm a gaming enthusiast, but I am just getting my feet wet with good hardware. My last PC that I just upgraded from was a Dell Inspiron 5675 with a Ryzen 7 1700 with a cooler worse than the stock cooler, so I definitely learned nothing about overclocking/underclocking or benchmarking at the time.
 

Phaaze88

Titan
Ambassador
How do I post screenshots to this forum?