Share your AIDA 64 cache and memory benchmark here

[Ph03n1X]

Yes they were. I mean if they weren't it shouldn't just affect L3 cache? I should be running into BSODS or other stability issues and my other scores would be bad too? I get the same performance as other in CPU-Z, TimeSpy, Cinebench R20. It just AIDA Cache test giving me this... My ram is in 2 and 4 slots (from CPU) like it supposed to be

Here is my timings:

I do not know enough to validate it. I had issue in past - one pin (not on CPU, but hole on AM4 socket) was damaged, and it made pcie-express x8 from original x16.
After fix, it was ok again.
So, if some pin is responsible for how ram communicates with CPU/L3... with on of lines is damaged, why it could note behave this way?
But if is something wrong with your CPU, I think it will be issue with infinity fabric between SOC or controller. And L3 will act the same in different PC.
But I hope not. I hope when you change your RAM, completely reset CMOS etc., you will have appropriate speeds.
Good luck.

 

[arabus]

 

[craxton]

What would cause the memory copy speed to be low here? I've looked around and other 3700x owners with a similar latency score, seem to be getting around 50k mem copy? Mine always sits around 45-46k max. I'm using the Ryzen high performance power plan and latest agesa 1.0.0.6 bios. Cinebench r20 scores seem ok at around 500 single core, and 4700 multi core.. Ram is Crucial Ballistix Sport 3200mhz overclocked to 3600mhz (using dram calculator timings).

View attachment 176442

you could fix that trfc they all shouldnt be the same....you can use ryzen dram in additional cal tab to get the correct values you need or pretty dang close to it for the three you should have set manually instead of setting the first and it setting all 3 to the same as zentimings is showing thats what you done here. i dont have the timings from my 1600af persay, can get them for my 3600xt to show comparison but the 5600x should give a good enough ex of what im talking about. personally trying to figure out why my l3 cache is so low. trying the edc bug in bios only so far limits my cpu from boosting to 4.85 ghz... NOTE: retest after you fix this. and your trfc in NS should be (what i used) is you XMP profile which you can find in bios under dimm info for XMP which isnt a setting its just info about the ram.

 

[jesdals]

Did some pretesting before new bios update testing

 

[craxton]

Did some pretesting before new bios update testing
View attachment 179632

Was your l3 cache always that range? 5600x is quite low considering my 4000mhz oc 2000 fclk and timings shared above. Compared to r5 3699xt 600 or so the l3 cache I feel somethings off. Beta patch d1 for my board is trash and locks ram to no post even at 3200c14 tforce ram.... Did try the beta bios and the results were worse hence lower ram oc. Also tried the edc bug mentioned in another forum but that didn't help only made my pc boot loop finally posted but still did nothing to help. Which is limiting pbo 150 pot 100/100 curve to negitave 20...anyone else experience low l3 cache with rather decent overclock on the ram and rightish timings????

 

[Akira1364]

Very glad I did not listen to the people who said "just buy the cheapest kit of DDR3-1600" back in 2013. Early last year I swapped out the GTX 780 the PC featured below originally had for a GTX 1660 Ti, and I'm fairly confident I'd be getting much worse results with it if I did have something like DDR3-1600.

 

[harm9963]

Waiting for new BIOS, hope soon

 

[craxton]

Waiting for new BIOS, hope soon

SAME as the L3 cache on my 5600x is TRASH compared to my 3600xt which should NOT be the case as my 3600xt only ran the ram at the exact same timings only 3800mhz 1900fclk to where as well...you can see..score is a little better without chrome running but still the l3 cache is well its just not what it should be from what ive seen with those on the new agesa release 1.1.8 i think. to where as msi hasnt released ANYTHING worthy of going to past patch C as D1 is WAY worse than C..or for me is was, couldn post XMP profile let alone 3200LOOSE timings at 1.5 volts. for ryzen 5000 series being the way they claimed its interleaved it should be nearly 2 times the speed it is now. which is killing performance

Unfortunately I dont know anyone with AMD. Most of my IRL friends are using laptops or older intel CPUs.

you dont need to try different ram. just increase frequency. also, the agesa atm for ALOT of motherboards is kill L3 cache speed which is wide spreading around the net like the plaque atm. so dont worry to much mines 4000mhz 16,15,15,15,32 timings with decent sub timings manually tuned with 2000fclk at that and its nowhere near what a 3600xt (my 3600xt) gets at 3800mhz 1to1 mode) on this same page above i have a shot of my 3600xt getting WAY better l3 cache speeds. some in other places have modified some code in the bios on their boards (their not sharing it tho) and have figured out whats the cause. it is not an aida thing causing this. as you can use a few other benchmarks to test it and itll show somethings off with agesa 1.1.0.0. so waiting for 1.1.8 or 1.1.0.8 unsure which its called without looking

that should solve l3 issues (hopes) but may lock out 2000fclk overclocks entirely without any option to allow users to pass thru it even thru LN2 mode inside the bios...anyhow yea dont worry about it. also use memory try it in your bios for a 3600 or 3800 overclock. you kit should be B-die with the 3200c14 timings so your kit in theory should do pretty close to the same timings and speed as my ram does. granted your processor allows it and motherboard maker as well.

Very glad I did not listen to the people who said "just buy the cheapest kit of DDR3-1600" back in 2013. Early last year I swapped out the GTX 780 the PC featured below originally had for a GTX 1660 Ti, and I'm fairly confident I'd be getting much worse results with it if I did have something like DDR3-1600.
View attachment 179663

your graphics card has nothing to do with how aida tests for cache and memory benchmarking lol. also, id have to guess you could essentially get more out of your ram granted your willing to have bios resets ect. unless your happy where you are currently.

 

[arabus]

 

[INSTG8R]

View attachment 179764

There is definitely something off about your IF clock(NB Clock)you may need to manually set your IF clock in the BIOS like I had to or else its starts using dividers like your result seems to show. With your fast RAM 1900 should be easily achievable

OC RAM(3600 to 3800) OC IF to 1900, Stock RAM timings

 

[Akira1364]

your graphics card has nothing to do with how aida tests for cache and memory benchmarking lol. also, id have to guess you could essentially get more out of your ram granted your willing to have bios resets ect. unless your happy where you are currently.

I was talking about the performance I get out of my modern GPU with the faster RAM versus what I'd get with slower RAM.

 

[INSTG8R]

Waiting for new BIOS, hope soon View attachment 179731

AIDA64 just got an update for better Zen3 support might fill in some of ”the blanks” in your bench(NB clock for example)I’ve got a 5600X ordered looking forward comparing it to my 3700X Using my same RAM a clocks/timings

 

[tufkal]

Hello all, very new to all of this and looking for some guidance. I'm putting together my first system w/ DDR4 that uses XMP, and coincidently my first Ryzen system as well.

Gigabyte B550M-DS3H
AMD Ryzen 5 3600
Kingston HyperX HX432C16FB3K2/32

I am brand new to XMP, and from what I gather the basics are this:
My memory is clocked at 2400Mhz stock, but a pre-defined profile called a XMP profile can be applied that "safely" overclocks the ram beyond the default JEDEC standard. (Correct me if I am wrong)

NOTE: I picked this RAM because it was from the QVL list @ 3200mhz and it had no RGB nonsense.

I had some troubles with memtest86 errors after doing a BIOS upgrade, and I think I have it narrowed down to Profile1 being unstable on newer BIOS versions with 5000 series support (this is a totally unrelated issue). For now I am using BIOS F3, which I get stable results using Profile1.

I want to make sure everything is working properly. I'm using the AIDA64 test, and I want to make sure the memory is performing as expected.

https://imgur.com/a/3CtpdGt

Are the speeds reported in line with what I should expect DDR4-3200? The write speed seems odd to me.. but maybe it's fine? I keep seeing AIDA64 screenshots showing a lot less latency and no dip in write speed, so I'm worried I'm doing something wrong. Other 3600 owners w/ PC-3200 out there?

 

[INSTG8R]

Hello all, very new to all of this and looking for some guidance. I'm putting together my first system w/ DDR4 that uses XMP, and coincidently my first Ryzen system as well.

Gigabyte B550M-DS3H
AMD Ryzen 5 3600
Kingston HyperX HX432C16FB3K2/32

I am brand new to XMP, and from what I gather the basics are this:
My memory is clocked at 2400Mhz stock, but a pre-defined profile called a XMP profile can be applied that "safely" overclocks the ram beyond the default JEDEC standard. (Correct me if I am wrong)

NOTE: I picked this RAM because it was from the QVL list @ 3200mhz and it had no RGB nonsense.

I had some troubles with memtest86 errors after doing a BIOS upgrade, and I think I have it narrowed down to Profile1 being unstable on newer BIOS versions with 5000 series support (this is a totally unrelated issue). For now I am using BIOS F3, which I get stable results using Profile1.

I want to make sure everything is working properly. I'm using the AIDA64 test, and I want to make sure the memory is performing as expected.

https://imgur.com/a/3CtpdGt

Are the speeds reported in line with what I should expect DDR4-3200? The write speed seems odd to me.. but maybe it's fine? I keep seeing AIDA64 screenshots showing a lot less latency and no dip in write speed, so I'm worried I'm doing something wrong. Other 3600 owners w/ PC-3200 out there?

Looks fine to me running at 3200, latency seems within reasonable numbers.

 

[tufkal]

Looks fine to me running at 3200, latency seems within reasonable numbers.

Thank you! That's exactly what I'm looking for. I keep finding all these ridiculous benchmarks for overclockers, but all I want to know is does that look normal for stock w/ XMP1.

 

[INSTG8R]

Thank you! That's exactly what I'm looking for. I keep finding all these ridiculous benchmarks for overclockers, but all I want to know is does that look normal for stock w/ XMP1.

Looks like it’s running the 3200 XMP profile just fine. There’s a big rabbit hole that OC’ers go down to get the crazy numbers.

 

[dont whant to set it"']

Darn stubborn motherboard is stubborn as it just wont set CommandRate to 2 , even tough I set it in both menus (dram timig page/amd cbs) as a try.

 

[maximoor]

R5 2600 @ 3.9 Undervolted to 1.23v & 2x8 RipjawsV 3200 Cl16

 

[miloaisdua]

is this any good? using patriot viper 4 blackout edition 2x32gb 3600 cl18 kit. followed some online guides for timing adjustments. can't seem to get fclock to go beyong 1866mhz.

Spoiler: settings

vddcr_soc=1.1
dram_v=1.43
vddg_ccd=1
vddg_iod=1
cldo_vddp=0.9
vttddr=0.715
procodt=60
cmd2t=1t
gear_down=enabled
power_down=enabled
rttnom_disabled
rttwr=auto
rttpark=rzq/4
cadbusclkdrv=24
cadbusaddrcmddrf=20
cadbuscsodt=24
cadbusckedrv=24
fclk=1866
mhz=3733
tcl=16
trcdrd=21
trcdwr=20
trp=20
tras=36
trc=62
trrds=4
trrdl=4
tfaw=16
twtrs=2
twtrl=7
twr=10
trcpage=0
trdrdscl=5
twrwrscl=5
trfc=560
trfc2=416
trfc4=256
tcwl=16
trtp=5
trdwr=6
twrrd=2
twrwrsc=1
twrwrsd=6
twrwrdd=6
trdrdscl=1
trdrdsd=4
trdrddd=4
tcke=1

 

[freeagent]

Just getting a feel for her.. no cooler yet so still a stock CPU.. PC hasn't shut off yet so that's promising lol

Stix are at 1.45v.. probably wont go any higher.. now I can start tightening things up I reckon.

 

[biffzinker]

Darn stubborn motherboard is stubborn as it just wont set CommandRate to 2

Gear down mode is on if the setting doesn’t stick.

 

[dont whant to set it"']

From what I read the in motherboard manuals, gear down mode is for untieing the primary timings so as to be able to set them oddly , like for example 14-16-15 or 13-14-15 ... and not much if nothing to do with command rate setting.

Anyhow, I swapped my dual rank dimm kit wich requires 2T timing for best performance for a single rank dimm kit wich is from among the first generations of commercial ddr4 rams, it dosnt clock to high , to about 2666MHz stable nor do serious tight timings , all in search of my CPU's IF clock capability.
Started at 1600MHz Dram with 10-10-10-20-40 1T with the IF clock of 800MHz , bumping up the IF clock ever so slightly till it reached 1866MHz , after witch I raised the RAM speed a bit for 1-2-1 ratio.

 

[biffzinker]

and not much if nothing to do with command rate setting.

Enabling Gear down mode overrides the command rate setting. To set a 1T command rate, gear down mode needs to be disabled.

@robot zombie?

 

[Adasiek74]

My Lenovo Y540: Crucial 32GB RAM DDR4 CL 19, GTX 1660Ti, Kioxia Exceria Plus 1TB + Crucial MX500 500GB

 

[dgianstefani]