BANANA Benchmark Results Comparison

If you think the dark rock 4 is huge, check out the dark rock 4 pro

 

Also depends from room temp as well as ventilation of the case etc. Here it is so warm that I should get lot less noise from fans when season changes, not sure about CPU temps though, precision boost gets low core load clocks quite high and then thing heats up, not using PBO though, which is Overdrive, that is overclocking feature, I did disabled that in bios.

I have mesh case and roof full of fans, but it is still lot of heat to get rid of, problem with setting AIO to roof would of been getting GPU heat trough AIO cooler which generally is not so great.

Good air cooler can do really well as long as hot air does not get trapped, but it is same for AIO, although my liquid temps don't get much hotter than +2C of ambient, this ambient is bit much for any proper gaming.

 

First place woohoo

 

I have the dark rock 3 and the pro is just massive in comparison. Also @fufsgfen i have a bequiet dark base 900 pro rev 2 wich is rather restrictive i also have PBO on to get a bit higher clocks since i have no clue of overclocking

 

Got an 8.1 install working well enough to run this. No monitoring software installed unfortunately. Specs are 8.1 Embedded, 5950x on water, 64GB 3600 cl16.

Stock Clocks no PBO or other motherboard trickery.



With motherboard limited PBO and curve optimizer enabled.



Could maybe get it to 850 if I spend some time on it but using windows 8 with no graphics drivers is rough so I may not.

 

There was a video I did watch about PBO and it made computer slower on some cases, it is lengthy, but about your CPU generation and quite thorough:


All I know is that indoor temperature here gets up by 2C with Ryzen just idling around and with ambient temps being above 30C that just starts to be bit much for comfort.

That pretty much confirms that Windows 8.1 scores higher than Win 10, also as of now Win 11 scores lot lower than win 10, that means these scores should probably be classified by OS, Would test Win 7, but that is such a trouble to get installed to these modern systems that might not be worth it.

Oh I did see a video where they did show how to install modern Radeon drivers to Win 8, all they did was that they put windows 7 drivers in by using compability mode (win 7) and then right clicking the inf file and installing drivers that way.
Also I have read comments by guys running rtx 3000 series of GPUs in Win 8, so I guess GPU drivers part might be possible to get sorted.


For me there is clear difference on how loading of stuff affects OS, same model SSD on both OS, so don't know why Win 10 gets huge slowdowns during heavy disk operations.

Also noticed that Bluetooth driver that is labeled as Win 10 at Asrock site, actually has Win 8.1 and Win 7 drivers, but those don't make Bluetooth to work in Windows 8.1, also Wlan is a challenge, but I don't use neither of those, so not a big issue, but yeah there are some problems for sure.

USB 3 ports work much better than with Z390 intel machine where file transfer stops and resumes constantly, so they work really slow, but with X570 and 5900X USB 3 works normally, I totally was excepting Win 8.1 to be unusable, but seems to be fine with my hardware. More issues might emerge with PCIe 4.0 as currectly I don't have anything that uses that, Samsung 980 Pro NVMe is waiting on desk though, but haven't decided what OS I will be running yet.

I might try to see if there is real practical measurable difference for running BeamNG as this score difference probably is just that and might not transfer to actual gameplay in any way.

Did you test Win 11 gameplay, was it any different to Win 10, despite score in Bananabench being lower?

 

Id actually like to bench win 8.1 vs 10 in a few benchmarks now maybe the bloatware in win 10 slows it down even more? Also with PBO ill try today if i have performance diffrences when its enabled and when its not.
Edit: I now ran banana bench and 3d Mark with and without PBO and both times it was faster but not by much also i noticed my temp are lower without PBO dont know why since it clocks higher without PBO. Also i´ll put the 3dmark and banana bench runs here too the first 2 are the ones with PBO the last 2 are the ones without

 

Look those numbers in Bananabench around 10 vehicles, surprising?

There is lot of variables in PBO that affect whole picture, so for example voltage with PBO is higher, which gives more heat and after CPU temp reaches certain level it cuts down the clocks.

On some situations PBO helps, especially when all cores are loaded as CPU temp is lower as voltages are lower, so there is more room to push CPU in regards of temperature, but then when only few cores are loaded, PBO might not give so much boost, but when one or two cores are loaded, there might not be penalty either.

Also I did read that PBO might contribute to CPU degradation, but I'm not sure how valid those claims are.

I guess that in Beam we might get actually hit of performance with PBO, but haven't done testing, it is only next week that I can do any proper testing at all, but for me to fully understand this stuff I need to do the testing and indeed difference of OS is something interesting, if that is only in Bananabench or if difference translates to actual gameplay.

Also clock speeds of at least 5900X are somewhat questionable. Without PBO I'm seeing 4950Mhz peak when playing BeamNG, but Ryzen master is not showing that kind of clock speeds, peak is 4800Mhz in there and quite rarely of that, also in new HWinfo64 there is mentioning about clock speed not being very accurate with modern CPUs, there is effective clock speed that matches much closer to what Ryzen master shows.

So what is then actually clock speed that CPU runs when doing banana bench, that is difficult to give any answer, because clock speeds change so wildly during testing, "effective clock speed" also never reaches the numbers "clock speed" reaches.

Point being, when quoting clock speed for bananabench, it is kind of impossible for me to give actual figure, there is too much of randomness and it is each core that is different and load jumps from core to another, nothing like Intel where everything stays put and it is super easy to point clock speed changes by load or settings etc.

So for spreadsheet that probably makes some challenges to get accurate statistics out of these new CPUs and there might be also some confusion in addition to different OS giving different results. Although it still probably is perfectly good for this purposes, it is good to know that there can be quite bit variation.

 

Clockspeedwise i normally use Task Manager and acccording to it the clock speed is in the 4000-4200 range. Also the jump at 10 cars is something i havent noticed when looking at the results. Another thing i ran banana bench around 10 times and basically always got hugely varying results of around 50 points on one car i dont know what the issue there was though. I also noticed my ram speed really affects the cpu alot so ryzen does really like fast ram my 3dmark score improved alot by just increasing the ram speed from 2666mhz when i originally got it to 3466mhz now (i have 3200mhz ram but it wouldnt run at that level at first because the bios wasnt Updated) the only thing id like to try now would be overclocking my cpu but I have no idea how to and dont want to risk my cpu because of it.

 

Spec :
Gigabyte A320M s2h
Ryzen 5 1600 with Stock Wraith Spire No LED cooler
single stick of Micron MTA8ATF1G64AZ-2G6E1 (8GB 2666MHz CL20)
MSI RX 460 4GB

No Overclock or anything on the cpu because the mobo doesnt let me to
ram is overclocked to 3533MHz CL18 Auto Voltage
CPU is boosting to stock all core boost clock (3,396MHz)
No PBO because i cant find the setting in the bios nor Ryzen Master so the cpu running at completly stock speed
Unfortunately dont have HWiNFO opened when doing this 30 cars bananabench to achieve score as high as possible but i do remember what speed its running when doing other bananabench
additional info : console.x64.exe priority set to Realtime

 

Mind you that I generally play the game at 4K UHD on maximum settings just fine, so I ain't complaining xD

 

I have also a 3900x with an Xilence 240mm AOI
on a MSI B450 gaming plus max Motherboard
32gb 2666mhz crucial RAM (oc to 3000)
RTX3080
2x 650w power supply

what could be the reason why my results differ after the 6th
cores so much ?
Maybe the Ram or the motherboard ?

something is botlemnecking me definitely

 

Decided to go run this again after a long time of not running it... you can see the bump up in physics performance from the left one (older 2019 build) to the right one (where we sit today). Thats a 26% increase in performance on the exact same computer. Pretty impressive dev's... pretty impressive.

 

One thing thats definatly not helping is the ram speed ryzen loves fast ram official sweet spot is supposed to be 3600mhz. Also check your thermals when you run banana bench maybe the aio isnt up to the task somehow?

 

i7-11700F @4.4GHz
---EDIT---
491.998 MBeams/s
516.260 MBeams/s...with disabled unnecessary applications and services

 

I most definitely cannot run 16 cars - my laptop starts struggling at 7. i5 9300h, 8gb ram

 

Yes, the total CPU performance during gameplay is consumed on multiple task like prepare graphics stuff for GPU, managing peripheral inputs(keyboard, mouse, wheel),...so not only calculating car physics like in benchmark.
But it is interesting to see, how much the number can drop.

 

Anyone got a 128 core thread ripper score ? Curious what it would be.

 

<https://docs.google.com/spreadsheet...EGg29QSORB1X_Zgs6KoFZ8LXQ/edit#gid=1183496459>
No, there are a couple of results of 1st gen Threadrippers, but they're not 128 cores.

 

I don't think a 128 core Threadripper exists yet - I think it only goes up to 64, but there are rumours of the next gen having 128