6400c30 vs 8000c36 Ryzen 9800X3D

[RootinTootinPootin]

On topic, 6400 offer a bit lower latency, 8000 offer a bit bigger bandwidth.

on AMD whether its 50 or 60ns latency, when gaming it feels "more", at quite some time in COD, the movement gets a little bit "sticky" in feel no matter what you press..unlike on Intel (I don't want to start another Intel vs AMD regarding this issue, this is based on my real world testing as I own all three latest platforms and all are tweaked to its hilt)

Some games benefit from higher bandwidth, others from lower latency and it's vice versa in different games/programs, and we're talking about few FPS, so under 0.5-1% difference that's fine for benchmarking but irrelevant for 24/7. Yes, when the benefit is for 0.1% Low, then that's nice.

based on my testing, 1:2 gives higher/better lows compared to 1:1, based on my testing, what impacts more is the sensitivity to temps, sudden frequency drops really matter alot specially on benching.

What's interesting is going too high will actually reduce stability etc. So there is a sweet spot. Usually between 1.2 and 1.3.

Silicon lottery in play, so pick your own poison..

 

[AVATARAT]

on AMD whether its 50 or 60ns latency, when gaming it feels "more", at quite some time in COD, the movement gets a little bit "sticky" in feel no matter what you press..unlike on Intel (I don't want to start another Intel vs AMD regarding this issue, this is based on my real world testing as I own all three latest platforms and all are tweaked to its hilt)

Intel's prediction is a bit better so in some situation, I agree. In fact they are so close to x3D without so much cache. In the end both platforms have their proc and cons.
You can try with SMT off (BIOS > Lasso), on some engines this works really flawless.

based on my testing, 1:2 gives higher/better lows compared to 1:1, based on my testing, what impacts more is the sensitivity to temps, sudden frequency drops really matter alot specially on benching.

Low's depend mainly on the engine from what I know.
Did you test with CapFrameX?

Silicon lottery in play, so pick your own poison..

Try lowering the vSoC if you use the latest AGESA, for me it running stable 0.075v less was a big surprise.

 

[RootinTootinPootin]

You can try with SMT off (BIOS > Lasso), on some engines this works really flawless.

I don't (or I won't) have time to do this much effort when I already paid enough for the price of the platform and just expect it to perform as advertised, Intel is easy very easy to deal with in terms of ease of use, I just dial my memory OC and CPU OC and it just works out right..

AM4 and AM5 really stressed the hell out of me just to get things running right on track (Dual CCD's, single CCD's aren't my thing as I need the extra horses for work and other stuff, non-X3D chips are also easy to tame, I just downgraded my son's gaming setup from an i7 14700KF+DDR5 7600 cl34 down to a Ryzen 7 9700X+B650E Tachyon+DDR5 6200 cl28+FCLK 2100, and he's not complaining much about stuff, specially with CS2, but overall he can tell the i7 has a better click and response latency)

Did you test with CapFrameX?

Yes and Presentmon too..

Try lowering the vSoC if you use the latest AGESA, for me it running stable 0.075v less was a big surprise.

Nahh!! I just slapped my ARL platform back in just to get rid of the headache every weekend, last week I was able to enjoy a total of 16hrs with my games (replaying The Surge 2), previous week I was never able to enjoy gaming with the Ryzen Platform as the dips really would kill my gaming mood, on ARL, I may not get the best maximum FPS on 4k but at least it never dips and the lows and frame pacing is very smooth, I'd really care less with my max FPS, and between those 2 platforms its just around 8-10 fps difference for me in 4k, not really astounding despite the Ryzen 9 being hailed a better chip vs the Ultra 9 285k.

 

[AVATARAT]

Yes, if you need more than 8 cores, Intel is somehow more interesting.

...I updated to the latest bios/agesa yesterday and tested it, before I needed 1.275v vSoC for FCLK 2200 (6200MT/s), now it can run with 1.2v (I need to test lower).
So AMD improved it, thanks for the info!

After more testing I up vSoC to 1.22v but I appreciate AMD's work (or maybe just Asrock), is really nice that they improve it to work with lower voltages.

 

[ir_cow]

If .75 is stable, I want to see what 1.3 can do in Gear 2 now. Time to load up some 9600 memory and go to the moon

 

[ShrimpBrime]

If .75 is stable, I want to see what 1.3 can do in Gear 2 now. Time to load up some 9600 memory and go to the moon

Hey, the OP spec lists this board.

Asrock B650m HDV

I would confirm because this board isn't going to go anywhere near 8000mt/s.

 

[ir_cow]

@ShrimpBrime I have my doubts of stability.

 

[ShrimpBrime]

@ShrimpBrime I have my doubts of stability.

I looked closer, Maybe the OP needs to update the system specs? I looked closer at the rudimentary screen shot and saw an ROG board and 9800X3D, I must be mistaken. My bad.

 

[RaceT3ch]

I have tested these settings vs each other and they perform quite similar. However 8000 is much better due to one thing: it requires far less soc voltage due to 2:1 so consumption, temps and noise on 8000 is far superior.

Okay, why shouldn't I just run the RAM at 4000 then?
Cause 2:1 at 8000 equals 4000MT/S

Besides, if you're worried about power usage, you'll find a bigger improvement by undervolting the cores instead of the SOC.

I'll also add, X3D takes 3 times fewer trips storing info from RAM into cache, as it has 3 times more L3 cache compared to the 9700X and non X3D counterparts. Meaning that X3D CPU's are less RAM sensitive.

 

[Cowboystrekk]

Okay, why shouldn't I just run the RAM at 4000 then?
Cause 2:1 at 8000 equals 4000MT/S

Besides, if you're worried about power usage, you'll find a bigger improvement by undervolting the cores instead of the SOC.

I'll also add, X3D takes 3 times fewer trips storing info from RAM into cache, as it has 3 times more L3 cache compared to the 9700X and non X3D counterparts. Meaning that X3D CPU's are less RAM sensitive.

You get general positive scaling up to around 6000, in a few cases above. I have of course undervolted cores. They sit at -31 co avg. x3ds cares less for ram, but as I showed in an earlier post you can still get 10%+ perf in some games/apps.

 

[ir_cow]

AGESA PI 1.2.0.3f. Did the opposite for me. Now it REQUIRES 1.3 SOC for 7000 Gear2. Spent the last day trying to figure out why y-cruncher would fail sometimes and not others. Narrow it down to the SoC after a bunch of memory kits at different voltages and speeds from 7000 to 8000 nothing but 1.3v SoC passed memory stability over a hour. Went back to - AGESA PI-1.2.0.3c and all my problems went away.

 

[Cowboystrekk]

AGESA PI 1.2.0.3f. Did the opposite for me. Now it REQUIRES 1.3 SOC for 7000 Gear2. Spent the last day trying to figure out why y-cruncher would fail sometimes and not others. Narrow it down to the SoC after a bunch of memory kits at different voltages and speeds from 7000 to 8000 nothing but 1.3v SoC passed memory stability over a hour. Went back to - AGESA PI-1.2.0.3c and all my problems went away.

What was soc req for 1.2.0.3c?

 

[ir_cow]

What was soc req for 1.2.0.3c?

1.2V if its in Gear 2 up to 7600. 6400 1:1 and 8000 2:1 needed 1.3V SoC before as well, but with AGESA PI 1.2.0.3f it isn't stable and will not pass y-cruncher 2.5B or a hour of memtest5.

I see ASUSS isn't allowing downgrades either. Beware ASUS users, you are stuck with 1.2.0.3f if you upgrade.