I ran it on my 5800X3D & RX 6800 XT. Looking at @freeagent score, I have room for improvement with the CPU multi core.
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
New Cinebench 2024
- Thread starter mtdewcowboy
- Start date
Just running my FCLK/MCLK combo at 1933 flat 15s. 255 tRFC, and -25 All Core.. And I was using my Grey FC140
Yours looks pretty good
- Joined
- Mar 28, 2003
Literally the only change here, now, is the motherboard. Now running the EVGA Z790 Dark K|ngp|n, coming from the EVGA Z690 Classified.
+529 GPU
+40 Multi Core CPU
+6 Single Core CPU
I am rather surprised at the results. I really didn't think the results would increase this much with everything at stock settings/speeds. Even the ambient temp was adjusted to match as closely as possible to the Z690 run.
+529 GPU
+40 Multi Core CPU
+6 Single Core CPU
I am rather surprised at the results. I really didn't think the results would increase this much with everything at stock settings/speeds. Even the ambient temp was adjusted to match as closely as possible to the Z690 run.
Attachments
- Joined
- Dec 18, 2000
Virtualization-Based Security (VBS) is a security feature in Windows 11 that creates an isolated region of memory for security features.
I am running Hynix RAM at 3733 16-19-19, so that is probably the difference. That B-die RAM is some amazing stuff! I tried going from 3733 to 3800 1:1 to improve things, & the board didn't like that & cleared CMOS to get going again. I hope I saved an OC profile on this BIOS version, otherwise I've got to manually retune everything.Just running my FCLK/MCLK combo at 1933 flat 15s. 255 tRFC, and -25 All Core.. And I was using my Grey FC140
Yours looks pretty good
I am running Hynix RAM at 3733 16-19-19, so that is probably the difference. That B-die RAM is some amazing stuff! I tried going from 3733 to 3800 1:1 to improve things, & the board didn't like that & cleared CMOS to get going again. I hope I saved an OC profile on this BIOS version, otherwise I've got to manually retune everything.
I don't know what about R24 but in previous versions, RAM was barely helping. At least the significant difference was between maybe 4-6 memory ratios and something like standard timings vs very tight.
I have 603 points in multi-core and 99 in single-core on the i5-1340P (Intel NUC) with 64GB DDR4-3200 CL20. I have it set to keep higher power limit, but will run again at the default settings.
Edit:
At default settings, the multi-core score is 630, so this is surprising, but it's around Threadripper 1920X and the single-core score is around Ryzen 5800X.
Last edited:
I dont think I saw much of a difference between 3200C14 and 3733C14.. maybe 50-100 points.. pretty much within variance. With 2024 I have no clue, I have only run it a few times. Interestingly, my 4070 Ti scores thousands of points more with my 5900X vs 58X3D. And 3DMark as well..
- Joined
- Jun 21, 2002
I think freeagent and JLK are pretty close in per core performance. JLK's score is 65% of freeagent's and he has 66% (.6 repeating if you want) of the cores. If anything I maybe would have expected the lower power and clocks to hinder more than the cache helped.
In some benchmarks, yes. In this one, not sure.
Freeagent also ran it on his 5800X3D in post #9 and he scored 920 multi-core vs my 5800X3D's score of 883. We are both running -25 on the CPU, but he is running RAM at a higher clock and tighter timings than me. I'm 4-5% slower & IMHO think the RAM could be a part of it.
Our RAM difference is more like DDR5-6000 CL34 and 7200 CL30, using your example.
I'm going to retest at 3600 XMP and 3733 tuned when I get a chance and see if there's a difference. Turns out I didn't save a profile with this BIOS version.
Sorry. I wasn't clear. I was talking about the RAM I have on hand and tested with. I can't really compare it though as they were on two different mobos.
Unless it's bandwidth, you won't find much much of a difference with such a small range.
I noticed an increase in all scores compared to my last complete system at stock run but now with 6400MT/S OC on the ram & FCLK of 2167. Used XMP timings from its 5600MT/S profile & applied it at 6400MT/S. A problem with the bios on this board is no option to disable GDM unlike my Gigabyte board. Doing that also creates a noticeable uplift in performance... but on the MSI board it is what it is with their current implementation of agesa 1.0.0.7c, GPU is still on stock with these scores.
Interesting to see this benchmark scaling with ram bandwidth & FCLK being upped. Otherwise everything else on on stock.
Interesting to see this benchmark scaling with ram bandwidth & FCLK being upped. Otherwise everything else on on stock.
Nice! Looks like it went up a few %.
When you said stock, you really meant stock with the ram at 4800 CL40. I would expect a couple of percent increase from that to 6400 CL30.... also wondering if the FCLK had anything to do with the increase. Can't tell what did what without testing each alone.
When you said stock, you really meant stock with the ram at 4800 CL40. I would expect a couple of percent increase from that to 6400 CL30.... also wondering if the FCLK had anything to do with the increase. Can't tell what did what without testing each alone.
Yeah, stock as in everything in bios on auto - that was my first post in this thread if you look back & No XMP profile. Which also means PBO is auto also in that earlier run. The FCLK on stock auto defaults to 2000, even at 4800MT/S. CPU-Z doesn't want to show it in any of the tabs, I think HWiNFO can show it, can run them again but currently testing an OC on the RAM atm with CL32 for 6400MT/S - which incidentally is pretty rare for Samsung B die ICs without blasting more than 1.40v for VDD through them.
However with the FCLK thing, on my Ryzen 7 system (7700, non X) it defaults to less than that, If I recall correctly, something like 1933 when booting at 4800MT/S.
Similar threads
