Share your AIDA 64 cache and memory benchmark here

[bersy]

 

[cloud_t]

Hello from Spain.

I recently bought an HP workstation on ebay and I did the Benchmarking tests with Passmark Cpu Performance. The memory results were disastrous and barely reached 34% score in overall comparison with the rest of users.

In the CPU tests the results were good, but in the 2D, Memory and HD sections they were nowhere near what was expected,

One of the obvious problems I have seen, is that neither Passmark, nor CPU-Z are able to read the SPD information from the memory sticks (8 x 32GB ECC Registered DDR4 2666 + 4 x 8GB ECC Registered DDR4 2666= 288 GB in total).

Yesterday I bought the Aida64 license and now everything has changed radically and I think the results are abnormally exaggerated compared to other users.

I would like to get your help to know if I am doing something wrong.

Thank you very much in advance for your help.

Best regards, Javier G.

View attachment 299593View attachment 299594

I reset my password just to reply to you. The answer is actually very simple: you are not doing anything wrong with your RAM, your scores are much higher because you're running much more memory channels than most users here... Because you're on a workstation/server platform with a Xeon CPU. Actually, you're running TWO CPUs on a dual-socket platform.

Put it into contrast: most users are posting their results from Desktop computers where the CPU only supports 2 channels ("Dual Channel") of memory. These are "consumer" GPUs and the systems where they sit don't support more channels, or more sockets for a second CPU to double those channels. The Xeons you're rocking support 6 channels for 12 total. And your system appears to be fully populated on those channels, hence why your arguably worse RAM is performing much better - you have more RAM splitting the work, more "wires" going to the CPU's memory controller, and more cores to process it. You are essentially "stripe-RAID"-ing RAM at 6 times what Dual Channel can do. The fact all channels are populated actually means you're doing "everything right". But my guess is whoever built this machine, built it well configured. After all, this machine probably cost over 25k when new in 2017.

Now, let me be a bit of an asshole and say: you bought an uber expensive system, being a Xeon Scalable on the Gold segment, from 2017. Your machine likely cost upwards of 2.5-3.5k by ebay prices. Nobody who spends that kind of cash would be posting here, unless they are either full of money and a completely naive as to what you bought (and what for!), or just making a post to gather likes or raise some eyebrows

 

[freeagent]

I am going to reinstall 11 right away, I have had a few bluescreens on a couple of CPUs tweaking mems.

 

[AVATARAT]

7700x@PBO
64GB@6200MHz

 

[opheen]

13600k \ Pro z690-A DDR4 \ 4x8GB Ripjaws V 14-15-8-14-28@4000 Gear1 CommandRate1T

 

[msimax]

 

[opheen]

i5 13600K 5.4P\E4.3\Cache4.8 VCORE 1.3v Set & 1.24v Load LLC5 \ Msi Pro z690-A DDR4 \ G.Skill Ripjaws V 2x16gb DR 16-16-8-16-32@4200 Gear1 CommandRate1T

VCCSA 1.335v VDDQtx 1.325v DRAM 1.61v

 

[freeagent]

 

[Tatili]

I am going to reinstall 11 right away, I have had a few bluescreens on a couple of CPUs tweaking mems.

View attachment 303182

Hi regarding 5800X3D tweaking mems, Win10 looks more reliable to me, was able to push my 4 Viper stick at 4000Mhz, but as I land in Wheas19 territory, was happy to fix rock stable Profile at 3933 GearON BCLK 101.8 Bus included CPU 4.629

I am going to reinstall 11 right away, I have had a few bluescreens on a couple of CPUs tweaking mems.

View attachment 303182

Now I found a way to GearOFF with AddrCmd 56, tried already stable at 3866, testing in progress for 3933 BCLK 101.8 Wheas free, not happy at 102.25

 

[opheen]

14-16-8-16-28@4200 Gear1 CommandRate2T. VCCSA 1.34v \ VDDQtx 1.325v \ DRAM-VDIMM 1.68v

 

[freeagent]

Taking the truck out for a spin

 

[opheen]

My wife's ITX 11900KF - Gigabyte z590i Vision D (RAM OC is a nightmare on this board) - 2 x 16GB DR Trident Z Royale Elite

 

[maksover]

10900KF DDR4 4700 CL16-17-17-36

 

[onemanhitsquad]

 

[chrcoluk]

Couple of interesting observations, I get 10ns lower latency on my test OS install vs my main install with Nvidia drivers, apps installed etc.

My 9900k has lower ram latency vs 13700k with same RAM at same timings and same memory clock speed. 3000CL16 on 9900k with looser secondary/tertiary is almost as low latency as 3200CL14 on my 13700k.

Some of the CAS/clock combo's in this thread I think I can only dream off, never had my B-die anywhere near them.

 

[Joe Bauers]

My 7950X burned out and i came back to the Dark Side.

 

[masterdeejay]

newer bios, some tweaking.
Samsung B die

 

[opheen]

i7 13700KF Msi MPG B760i Edge DDR4 Wifi (M-ITX) Corsair Vengeance RGB RT 4600C18 ver3.44 2x16GB SingleRank Micron Rev:B (MT40A2G8VA-062E:B)

18-22-21-42@4266 1.52v Gear1 CommandRate1T

 

[xcazy]

Non-K 13th Gen @ A-die 6000 @7400

 

[masterdeejay]

Cheapest 16gb modules (15 usd each) but ECC REG so not all motherboards can handle it.
4x DDR4-2133mhz 2Rx4 16Gb Samsung old 2015 E-die.
@3200mhz 16-18-18-38-CR2 1.35v
Temps: 40-45C at load.

The 16 USD/stick Hynix 16gb DJR ECC REG 3200 1Rx4 modules at 4000mhz Stock CAS22 settings but 1.35v
I think this a price/performance winner for x299 platform. It is getting near to my Samsung B die kits.
6 modules, but I will buy another 2. (8 modules needed for optimal performance, 2 rank per channel)
Some say these DJR can do over 5000 but my cpu IMC cant handle more than 4000-4200
Temperature is 45-50 at load

 

[Tatili]

Cheapest 16gb modules (15 usd each) but ECC REG so not all motherboards can handle it.
4x DDR4-2133mhz 2Rx4 16Gb Samsung old 2015 E-die.
@3200mhz 16-18-18-38-CR2 1.35v
Temps: 40-45C at load.

View attachment 306476
View attachment 306477

The 16 USD/stick Hynix 16gb DJR ECC REG 3200 1Rx4 modules at 4000mhz Stock CAS22 settings but 1.35v
I think this a price/performance winner for x299 platform. It is getting near to my Samsung B die kits.
6 modules, but I will buy another 2. (8 modules needed for optimal performance, 2 rank per channel)
Some say these DJR can do over 5000 but my cpu IMC cant handle more than 4000-4200
Temperature is 45-50 at load
View attachment 306553
View attachment 306554

Hi enthusiast, your Read Write Copy speed are way way too high, not real, remind me to my old 5800X strange behavoir due to my RGB software installed!
I was running same weird speeds after installing SignalRGB software and running in background by default, sadly Aida reports couple ten of thousands more speeds value than real, so better check some of your Software running in Task Manager that are bloating your real results, nice job for your sticks BTW

 

[masterdeejay]

Hi enthusiast, your Read Write Copy speed are way way too high, not real, remind me to my old 5800X strange behavoir due to my RGB software installed!
I was running same weird speeds after installing SignalRGB software and running in background by default, sadly Aida reports couple ten of thousands more speeds value than real, so better check some of your Software running in Task Manager that are bloating your real results, nice job for your sticks BTW

It is normal speed for a quad channel.

 

[opheen]

i7-13700KF Msi MPG B760i Edge DDR4 Wifi (M-ITX) Corsair Vengeance RT 4600C18 2x16gb SinglRank .

17-23-8-23-43@4266 Gear1 CommandRate1T DRAM\Vdimm 1.58v VCCSA 1.345v VDDQtx 1.32v

 

[Det0x]

God bin 7950x3d

1:1 mode
6600MT/s 2200FCLK 28-37-32-30-62 GDM disable
~313 mb/s in karhu

Full screenshot with all voltages/hwinfo open:


2:1 mode
8000MT/s 2200FCLK 34-45-42-40-82 GDM disable
332.5 mb/s in karhu

Full screenshot with all voltages/hwinfo open:

Settings above completed 25k Karhu and 1.5 hours Y-cruncher 8.1

 

[opheen]

16-21-8-19-40@4200Gear1CommandRate1T 1.53v DRAM 1.345v VCCSA 1.3v VDDQtx 5.3P(2cores boost to 5.4) 4.2E 4.8Cache.

2x16GB SingleRank Corsair Vengeance RT 4600C18

Karhu 10000%

TM5 Anta777 Extreme

Y-Cruncher 2.5B loop 30Min