Topic: Benchmark is down (Read 2076 times)

tkolsto · « **on:** June 05, 2023, 03:08:29 pm »

benchmark is way down. around 6200 points, normally I get around 8500/8600. I have done the taming of defender. I also just added 2 x 8 gb ram, now total 32 gb ram and the ram is exactly the same bought in the same web shop. same type number.

log file is in the attachment.

I usually get this benchmark score below:

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 1,647 seconds
Single-threaded floating point math... 2,152 seconds
Multi-threaded integer math... 0,287 seconds
Multi-threaded mixed math... 0,352 seconds
Score: 4281

Running 'Image' benchmark...
Image creation / destruction... 0,152 seconds
Flood filling... 0,172 seconds
Direct copying... 0,263 seconds
Small renders... 0,482 seconds
Bilinear rendering... 0,219 seconds
Bicubic rendering... 0,817 seconds
Score: 10449

Running 'Database' benchmark...
Create database... 0,051 seconds
Populate database... 0,397 seconds
Save database... 0,065 seconds
Reload database... 0,024 seconds
Search database... 0,671 seconds
Sort database... 0,357 seconds
Group database... 0,370 seconds
Score: 11108

JRMark (version 30.0.40 64 bit): 8613

What is making it so poor score?

JimH · « **Reply #1 on:** June 05, 2023, 03:12:45 pm »

Try rebooting.

tkolsto · « **Reply #2 on:** June 05, 2023, 03:24:50 pm »

Tried that Jim.

Last time it was that my ram was not set to 3200hz if I remember correctly. Normally I just havet to choose xmp 2.0 profile 1 in bios and save and exit and reboot. But when I check this in manager the ram is not changed to 3200 but are set to default 2133 hz. So I then choose directly the 3200 in bios setting and then the ram shows 3200 hz in manger but still I get the same benchmark score. could it be when I choose the hz directly force the ram to be set to 3200hz in bios, maybe something more should be activated when one do it this way maybe?

tkolsto · « **Reply #3 on:** June 05, 2023, 03:41:52 pm »

delete

Manfred · « **Reply #4 on:** June 06, 2023, 08:49:46 am »

I had the same problem. But after the upgrade to MC31.0.17 I first did not recognize that JRWorker was running in the background and consumes ~90-100% CPU. After JRWorker was finished old results was back.

tkolsto · « **Reply #5 on:** June 06, 2023, 11:31:35 am »

Thanks for responding Manfred! My problem is the ram which I added another 16 gb to 32 gb and the ram is the very same type to the millimeter. now my setup is 4 x 8 gb. but bios is does not work as it should. I choose xmp profile(there is only one profile) and the setting above shows the cpu and ram which changes to 3200 mhz. but exit save reboot..when in windows this has not changed. I dont know to much about why this is so...bios accept the change..but when booting into windows ram is default. googled a bit...but I dont find any solutions. I have send a mail to asrock, but no response yet.

Maybe a reset of bios? or maybe that should be avoided. I have removed cmos battery before..but I think the bios is just deleting the settings. does not fully reset back to the bios version it came with, which is probably a good thing,because that could cause problems. maybe a reset could be it?

Awesome Donkey · « **Reply #6 on:** June 06, 2023, 12:30:31 pm »

You're running 4 RAM sticks (quad channel)? 4 X 8GB? You might have better luck getting a 2 X 16GB kit and running it in dual channel versus 4 X 8GB in quad channel (which running in quad channel is usually more taxing on the system and *can* lead to more system instabilities) especially if you're using XMP on Intel or whatever it is AMD calls theirs now.

tkolsto · « **Reply #7 on:** June 06, 2023, 12:48:02 pm »

Yes I read now on google that 4 sticks is not ideal, but I had these in my other pc and wanted to try this out. I had an extra 8 gb ram for the other pc. So I thought that would not be any problem. xmp is probably only for 2 stick. or maybe I should take all 4 out again and reset bios and try again in xmp mode. and if that aint working then choose xmp mode again and slightly higher voltage( I dont expect anyone to recommend the voltage tweak..but maybe take all 4 out and reset bios?

tkolsto · « **Reply #8 on:** June 06, 2023, 12:58:06 pm »

just checked some more on the ram and they all are single and not dual. some ram is dual but not the case for my ram.

tkolsto · « **Reply #9 on:** June 06, 2023, 01:49:58 pm »

took out all the ram and put them back in, resetted the bios. xmp are chosable as before but in windows only default mhz. So I changed xmp to auto and chose 3200 mhz and altered the voltage from 1.350 to 1400. also updating to latest jriver build ..0017. but still same poor benchmark and the ram mhz is 3200. I forgot to mention that earlier when I chose mhz directly and got it to 3200..the first benchmark was much better ..around 8200 ..but not 8500/8600 and the when I run the second benchmark it was back to around 6200 mhz.

cpu is intel i5 12600kf

tkolsto · « **Reply #10 on:** June 06, 2023, 02:21:01 pm »

Now I did a mem speed check in command prompt and there the ram speed said 2130 mhz and not 3200 mhz thus poor jrvier benchmark. But in task manager when clicking on ram it states that it is 3200 mhz. funny..look like 3200 mhz is not do-able without some tweaking.

Awesome Donkey · « **Reply #11 on:** June 06, 2023, 02:32:21 pm »

2133Hz is the base clock frequency for DDR4 RAM, anything beyond that (e.g. 3200Hz) is actually an overclock via the XMP profile. You're not mixing several different RAM kits that run at different frequencies and have different timings, are you? If they're not exactly the same in capacity, frequency and timings, trying to run XMP with it will likely cause issues like running at the max frequency that all RAM sticks support. For example if you're using RAM that only supports up to 2600Hz and you add a 3200Hz kit in, it'll all only run at 2600Hz max.

In my own experience, using a dual channel kit (2 X 8GB, 2 X 16GB, etc.) works best without having issues using XMP or potential issues from running at quad channel.

tkolsto · « **Reply #12 on:** June 06, 2023, 03:54:12 pm »

no the ram is the same an ordered from the same art. nr at store also type description is the same so they have not changed it..anyways I took them out and set everything back to what it was using xmp profile...but now it is weird. I get the same low score about 6200 and sometimes 7000 and sometimes around 8000...when i do benchmark one after another...but mostly I now get around 6200.

tkolsto · « **Reply #13 on:** June 06, 2023, 04:31:00 pm »

something with the cpu performance that made the poor benchmark...I installed the motherboard utility program and set cpu to performance..now I get even performance about 8400...several times now...little bit lower but that could be to jriver not being stable build..I have notice such small insigificant benchmark score before..so now everything is as it should be

update : this was shortlived..after a short while the score was lower again.

tkolsto · « **Reply #14 on:** June 06, 2023, 05:27:31 pm »

installed to 32 gb..look that task manager was showing right mhz and that cpu was at some savingmode that I corrected by mb utility program. now all looks good and get around 8400 score...and also this is a bit faster now from the little I have done opening programs and such...so I think it is a faster pc now....programs loading in faster..I think my pc just become a bit quicker as I now started to use it ...and not focusing on ram issue.which is resolved by choosing 3200 mhz under dram tweak in bios..this cmd test I think is not showing the correct mhz..
also cpuid is showing...

but now as I write this...in cpuid which loaded a bit quicker ...under SPD on can select each ram and it is not the same type dram manufactor is samsung and the other set is Nanya technology with higher cas latency around 24 and samsung around 14 /15 but in xmp mode both are have cas latency at 16 and part number is the same on both modules.. so I think it is ok because pc is a bit quicker/snappier. Time will show if it is stable.

update: uninstalled mb utility. benchmark still ok. maybe mb utility corrected something, but have to set powersetting in windows to high to get normal benchmark score.

tkolsto · « **Reply #15 on:** June 07, 2023, 12:34:44 pm »

Ok now intel i5 12600kf specs says:
-----------------------------------------------------------
Memory Specifications
Max Memory Size (dependent on memory type)
128 GB

Memory Types
Up to DDR5 4800 MT/s
Up to DDR4 3200 MT/s

Max # of Memory Channels
2

Max Memory Bandwidth
76.8 GB/s
----------------------------------------------------

max of memory channels is 2. so it is listed as support of only 2 channels.

But why does it get 32 gb ram in windows and also pc is a bit faster at least loading of programs and also in general I feel it is more snappy and I do not think it is placebo effect.

But I think I am buying a new set of two modules in the future.

tkolsto · « **Reply #16 on:** June 08, 2023, 02:52:50 pm »

forget about it being faster and snappier..I now changed back to 16 gb and it is not slower...loading may be quicker but that is all. benchmark was not fixed earlier either it ended up with the same low score 6200...
now I have it as I had it previous..
log2 file attached
is is possibel to see in the log where the problem lies?

tkolsto · « **Reply #17 on:** June 08, 2023, 03:20:47 pm »

when I reset bios and choose xmp like I had before...when I enter windows and start jriver the two first benchmark give 8400 score but then it fall back again to 6200...and math benchmark is always 1800 or so...

I have not meddle with anything but in bios other than fanspeed and xmp and direct choosing of ram mhz and this ram upgrade...

maybe windows reinstall it the only way now.

newsposter · « **Reply #18 on:** June 09, 2023, 11:39:57 am »

What is the factory rated speed of the ram you are using? Are you trying to overclock it via bios settings?

tkolsto · « **Reply #19 on:** June 09, 2023, 03:10:09 pm »

Well it said 3200 mzh on the web store but default is 2133 mhz. all I am doing is using xmp profil which is what is recommended as stable 3200 mhz from what I understand.

in specs it said

ram performance: 3200 mhz

and at bottom of the specs is said:

SPD - 2133 MHz - 1.2 V - CL15 - 15-15-15-36
Tested - 3200 MHz - 1.35 V - CL16 - 16-18-18-36

Awesome Donkey · « **Reply #20 on:** June 09, 2023, 07:45:46 pm »

Anything over 2133 MHz for DDR4 is an overclock, via XMP for Intel CPUs or DOCP for AMD CPUs. 2133 MHz is the "base" frequency for DDR4.

Some videos on DDR4 and XMP/DOCP:

https://www.youtube.com/watch?v=ZA47U8hAD_E
https://www.youtube.com/watch?v=WDzrq5IxBX4

tkolsto · « **Reply #21 on:** June 13, 2023, 11:25:38 am »

Reinstalled windows 10 pro...benchmark is almost there...I get about 7900 / 8200. But I used to get a steady 8500/8600? It is not that much but strange that it is changed. somethings is changed. even erased the ssd disk with killdisk.

and my ram now is set to default. but i never had any problems with xmp earlier. but why the score is not as it used to be is beyond me.

tkolsto · « **Reply #22 on:** June 13, 2023, 12:18:44 pm »

I now set the ram to xmp profile and also let iobit driverbooster do updates. after this it got better..now math benchmark is ok and image benchmark is ok...only database benchmark is doing poorer as ..I used to get around 11300 - 10900...but now is around 9900/10100..

what does database benchmark monitor? what parts of the machine does this benchmark poorer? why do this part perform poorer?

and I am testing it on jrvier 30 latest version..0096. and the same with jriver 31 version..017.

tkolsto · « **Reply #23 on:** June 13, 2023, 12:37:16 pm »

Also I noticed that if I do not close edge browser and I restart the pc..it is still open with all its tabs open. ...why does it not null it out after restart? the opera browser is not behaving like this. I might add the edge browser it set to standard in windows.
both browser is deactivated in windows task manager. also the option in windows under accounts that restart app after restart is also disabled. still it happens.

Awesome Donkey · « **Reply #24 on:** June 13, 2023, 01:14:25 pm »

I mean, does it really matter that the benchmark results fluctuate a bit? Unless you notice some actual performance issue and/or actual difference, and MC works fine for you regardless, I don't really see how it matters.

Even synthetic benchmarks like Cinebench, Geekbench, 3DMark, etc. can fluctuate a bit with the results between runs.

tkolsto · « **Reply #25 on:** June 13, 2023, 01:41:30 pm »

solved the edge browser reapearing after restart..was that some settings needed to be turned off under edge setting under system and performance...quick start turned off.

Yes I agree..this could be changing as new versions are made..but normally when last version is done (jriver 30) this does not happen. so it is something that does this to get low score on database score in the benchmark. Yes maybe not pushing this anymore now. But this is not fluctuation between runs though. Thanks for helping. learned something about ram from You thumbs up:)

eve · « **Reply #26 on:** June 16, 2023, 01:31:41 pm »

Quote from: tkolsto on June 13, 2023, 12:18:44 pm

I now set the ram to xmp profile and also let iobit driverbooster do updates. after this it got better..now math benchmark is ok and image benchmark is ok...only database benchmark is doing poorer as ..I used to get around 11300 - 10900...but now is around 9900/10100..

what does database benchmark monitor? what parts of the machine does this benchmark poorer? why do this part perform poorer?

and I am testing it on jrvier 30 latest version..0096. and the same with jriver 31 version..017.

You should very much reconsider using things like DriverBooster

Too much thread to read but,

If you're consistently 'losing' your XMP settings, there can be a few causes. The first one is obviously stability, boards aren't that dumb anymore, if something isn't 'looking' right during the POST or there were issues previously, they can often attempt to revert back to JEDEC speeds.

The other can be your CMOS battery. Seriously, they do die. (On the flip side, I've pulled out machines from nearly 20 years ago that were in service for a serious amount of time and their CMOS battery has been a-ok)

Intel XMP is a weird one, it at least used to be 'technically' a violation of warranty despite Intel pushing it pretty heavily.

It's unlikely to be a bad stick (since you don't seem to be experiencing massive instability) but you can throw memtest at your RAM for a night.
Also while I'm not sure what exactly JRiver's benchmark does for database perf, the database structure of JRiver is decently efficent, I've had *very* large music libraries that aren't that unwieldy.
When I used JRiver for music like 5+ years ago? Chugged along just fine with a few million tracks. Unless you're a serious hoarder, you're probably not dealing with a database that's going to push things. Meanwhile, the software I use for music now? Absolutely terrible database performance, like laughably bad.

Manfred · « **Reply #27 on:** June 16, 2023, 02:48:20 pm »

I have now the same problem: Benchmark was ~7100 for my Ryzen 7 5700G and is now 6656.
I upgraded from 16 GB RAM to 32 GB both Corsair 3200 DDR4. I also did a fresh install of Win11. I must check if the timings for the 16 GB=2x8 and 32=2x16 modules are the same.
I had to do a BIOS Update to get the new RAM working. MB is Asus ROG Strix B550-I Gaming.

Reason seems to be:
In Asus Armoury Create the mem frequency is only 2133Mhz but xmp setting is 3200.

I will investigate.

tkolsto · « **Reply #28 on:** June 16, 2023, 05:47:16 pm »

Manfred the difference You get is excactly the same difference in score that I get. about 500/450 lower. I bet if You got the low score on the last part of the benchmark..the database benchmark part..that is the last part...and it score about 1000 point lower than it used to be...the other two parts of the benchmar..math and ..image benchmark are ok..

I used to get around 11 000 give or take 100 or 200 houndred points. now I get around 10 100- 9900.

tkolsto · « **Reply #29 on:** June 16, 2023, 06:31:00 pm »

eve ...I have tested the ram earlier..no problems..mobo is about a year old..but I will check the battery..

About mhz...I thought initially I did not get 3200 mhz..but I do...but it does not show when checking using command promt. so I got confused and thought somethings was wrong. everything looks like it should. task manager shows 3200 mhz. I have always used xmp ..no issues..and driverbooster too...after I let it update drivers..things got better. ..then I uninstalled it...no resedues in program folder..

also erlier I also got low score (another thread I started) when I got low score..my ram was set to default...and when I used xmp...the benchmark score was back on track. about 8500/8600..even sometimes 8700. Now this time I dont know what it is..but Manfred is experiencing what looks to be the same..same deviation about 450/500 points. bet it is the database part in his case too..

changed battery..doubt the one was bad though.

also another thing..my lg oled b1..is keeping changing hdmi mode from standard to pc mode...I have unplugged quite a bit..but I am not so sure thats it...sometimes it does noting and sometimes it do the change..it seems..dont see an excact trigger that does it..have to see if it settles.

Manfred · « **Reply #30 on:** June 17, 2023, 04:14:06 am »

Quote

I will investigate.

I got the 3200 MHz back. I had to choose the D.O.C.P profile under AI Tweaker in the BIOS and set manually the RAM frequency to 3200. (I though as I bought the motherboard it sets it automatically - but may be my personal memory is wrong

)
Original benchmark before I upgraded to 32 GB RAM and install Win11 new:

AMD Ryzen 7 5700G 16 GB DDR4-3200 High POWER Mode
=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 1,422 seconds
Single-threaded floating point math... 2,096 seconds
Multi-threaded integer math... 0,186 seconds
Multi-threaded mixed math... 0,281 seconds
Score: 4767

Running 'Image' benchmark...
Image creation / destruction... 0,230 seconds
Flood filling... 0,202 seconds
Direct copying... 0,434 seconds
Small renders... 0,623 seconds
Bilinear rendering... 0,231 seconds
Bicubic rendering... 1,152 seconds
Score: 7659

Running 'Database' benchmark...
Create database... 0,049 seconds
Populate database... 0,626 seconds
Save database... 0,108 seconds
Reload database... 0,035 seconds
Search database... 0,748 seconds
Sort database... 0,416 seconds
Group database... 0,489 seconds
Score: 8702

JRMark (version 31.0.21 64 bit): 7043

New: AMD Ryzen 7 5700G 32 GB DDR4-3200 Balanced POWER mode
Running 'Math' benchmark...
Single-threaded integer math... 1,540 seconds
Single-threaded floating point math... 2,272 seconds
Multi-threaded integer math... 0,194 seconds
Multi-threaded mixed math... 0,293 seconds
Score: 4419

Running 'Image' benchmark...
Image creation / destruction... 0,246 seconds
Flood filling... 0,204 seconds
Direct copying... 0,449 seconds
Small renders... 0,672 seconds
Bilinear rendering... 0,253 seconds
Bicubic rendering... 1,238 seconds
Score: 7186

Running 'Database' benchmark...
Create database... 0,054 seconds
Populate database... 0,582 seconds
Save database... 0,115 seconds
Reload database... 0,050 seconds
Search database... 0,795 seconds
Sort database... 0,453 seconds
Group database... 0,436 seconds
Score: 8650

JRMark (version 31.0.21 64 bit): 6752

Differences are fragmented. My file server has an intel i3. The numbers have not changed there.
I can live with the difference.

Got it back: Set Energy Modus to high power - Default in Win11 is balanced:
AMD Ryzen 7 5700G 32 GB DDR4-3200 High POWER Mode
=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 1,536 seconds
Single-threaded floating point math... 2,265 seconds
Multi-threaded integer math... 0,193 seconds
Multi-threaded mixed math... 0,293 seconds
Score: 4432

Running 'Image' benchmark...
Image creation / destruction... 0,224 seconds
Flood filling... 0,174 seconds
Direct copying... 0,403 seconds
Small renders... 0,669 seconds
Bilinear rendering... 0,243 seconds
Bicubic rendering... 1,234 seconds
Score: 7464

Running 'Database' benchmark...
Create database... 0,055 seconds
Populate database... 0,560 seconds
Save database... 0,109 seconds
Reload database... 0,050 seconds
Search database... 0,785 seconds
Sort database... 0,449 seconds
Group database... 0,417 seconds
Score: 8868

JRMark (version 31.0.21 64 bit): 6921

tkolsto · « **Reply #31 on:** June 17, 2023, 08:34:09 am »

Yes it is true there is difference from time to time..have always been like that...glad Your fixed your problem...yes high power and set ram to 3200 mhz..will do it..

Ram fix did the trick for me last time...also before I lost my good score...I did not even have to use high power...just let it set on balanced and JRiver/Nvidia take what ever power that is needed...

and Your math benchmark is very high I guess this is the cpu test 4767(did you overclock cpu or something?) is very high...I never get more than 4300 regurarly 4260 ish. ..And You got an amd ryzen 5700g and I got e newer intel i5 12600kf...which I though would crush your cpu..but it is the other way around...and a much cheaper cpu too.. about 100 dollar cheaper...got dam intel sucks.

tkolsto · « **Reply #32 on:** June 20, 2023, 04:39:59 pm »

Manfred...Your math benchmark is very high I guess this is the cpu test part of the jriver benchmark..math benchmark is 4767(did you overclock cpu or something?)

Manfred · « **Reply #33 on:** June 21, 2023, 11:41:16 am »

No overclocking! No Tweaking!

tkolsto · « **Reply #34 on:** June 21, 2023, 04:21:33 pm »

You realize Your first part of the JRiver benchmark when you were at 16 gb RAM was very fast? 4767 points. I get only around 4250 points.

Funny that your score goes down when you upgraded to 32 gb ram. then your score is 4432 which is also very good.

Finally...do you experience any difference while using you pc now after ram upgrade?

JimH · « **Reply #35 on:** June 21, 2023, 05:46:03 pm »

The benchmark will always vary a little. It can also change a little when we make internal changes.

Manfred · « **Reply #36 on:** June 22, 2023, 06:25:38 am »

Finally...do you experience any difference while using you PC now after ram upgrade?

I have not upgraded only the RAM, its a transition to a new solution based on the old components with RAM Upgrade + additional SSD. I follow now the JRiver ID Architecture Pattern "PC=Workstation" + Media Renderer. DIY File Server is for static video content and Backup.

Why I am doing this? The File Server has Win 10 on an headless Asus P10I-S mainboard and is a dead end. Remote Access via IPMI has been dropped (IE based) and the new Asus P12I does only support server grade OS no Win11!

Before I had my Workstation and a Media server (MC managed via RDP + all media content: Audio, Documents, Images, Video) ->2 MC Instances.

Now I have only one MC instance on the workstation with additional SSD for all new video content in 2023 and the old SSD moved to the workstation for Audio, Images, Documents.

The File Server has than all "old" < 2023 staic video content on 38 TB HDD + 8 TB Backup Space for Workstation backup. I have moved Acronis Backup to the workstation. In 2025 (EoL of Win10) File Server will be a NAS or Thunderbolt NAS.

So far up to now its stable. Running Acronis Backup total RAM consumption is sometimes > 16 GB . That's the reason why I upgraded and having local MC instance instead of using MS RDP makes a much better user experience.

tkolsto · « **Reply #37 on:** June 22, 2023, 09:51:58 am »

Yes Jim Thank You for writing this. Yes I know this and have never reacted on this. I used to have 8500/8600 and now have about 8100 /8250. and it is mostly the low score of the last part of jriver that gives this low score as it is about 1000 points lower. I do not know what this last part is about other than its name ..database benchmark...

further I do not believe that there is anything wrong about jriver..but my low score now is due to something but I do not know what. Also I Am not going to do anything more about this as I do not think I will ever find it out.

INTERACT FORUM

Author Topic: Benchmark is down (Read 2076 times)