So, hard drives DO affect in-engine performance

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
Running Unigine Heaven 4 currently

C drive scores are 850 (WD Black 500gb)
D drive scores are 830 (WD Black 3tb) with noticeable stuttering in the benchmark

Maybe there is a setting on the hard drives that can affect this? Is there something I should check for?
 

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
great, now I am getting 820 on the C drive as well.

I was just getting 851 consecutively... installed radeonpro, and moved Unigine over to the D drive, and now BOTH hard drives are performing at the weaker D drive level
 

Kippa

Senior member
Dec 12, 2011
392
1
81
Has anyone who has a lot of ram tried installing Unigine Heaven 4 on a Ram disk and seeing if there is a big performance gain? I'd like to see the result of that test.
 

BFG10K

Lifer
Aug 14, 2000
22,709
3,000
126
850, 830, 820...all irrelevant. You're within a ~3% margin of error, so it's nothing more than benchmarking noise.

A modern PMT/HAL operating system is always going to have slight variances because of other things happening at the same time, and/or slight configuration differences.

If you want precise timing then go use a hard-coded embedded system that can only do one thing.
 

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
it is not irrelevent

this system scored between 850-853 for the entire time its been setup

now all the sudden i install a 2nd hd and performance is lower than it should be

unigine scores should not drop like that, and i can see visible hitching while the benchmark is running that is not there before
 

AdamK47

Lifer
Oct 9, 1999
15,653
3,518
136
Unigine benchmarks will access the drives to stream in content. There are a few points in Heaven and Valley where this is consistent.
 

Railgun

Golden Member
Mar 27, 2010
1,289
2
81
Why are we assuming it's the drive as opposed to you're now accessing a remote disk for all intents and purposes. Could be a controller issue...could be a latency issue from the application and the delay pulling data from D. Does any data from D need to get into C? The OS is now pulling data from somewhere else instead of natively within its own disk.

Just tossing some additional scenarios out as I don't know the ins and outs of the app. I doubt it's the drive itself.
 
Last edited:

Kippa

Senior member
Dec 12, 2011
392
1
81
If you are using hard disk drives what about fragmentation? Any of your drives relatively full or have been used a lot over a long period of time.
 

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
Unigine benchmarks will access the drives to stream in content. There are a few points in Heaven and Valley where this is consistent.

yes and all the access points are extra stuttery right now for some reason, it brings my min fps way down
 

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
is not a fragmentation they are both clean hard drives

it feels like a data transfer issue though i have no idea what could be, the telltale hitching during unigine heaven looks like its not moving some data fast enough or like a latency issue

its not even the drop from 850 to 820-830 in the Unigine score that concerns me most, its that my MIN Frames are going from 18-19 down to 5 causing massive stuttering
 

therealnickdanger

Senior member
Oct 26, 2005
987
2
0
1. Not all hard drives are created equal when it comes to seek times. On top of that, some hard drives have super aggressive power management and will park the read/write heads more often than others, resulting in even worse seek times and overall bandwidth.

2. Defragging a hard drive is a bit of a misnomer. The process in Windows defragments individual files so that all the bits of ONE particular file are physically stored in only one area, but it doesn't magically group multiple files together (i.e.: all Unigine files in one sector), so the head still has to access multiple locations on the disk to run programs and load game assets. To optimize your HDD beyond what Windows defrag does, you need another program.

3. BFG is correct that these scores are within an acceptable margin of error. Every new HDD you add, Windows will adjust its virtual memory pool (page file) for each drive. You could try disabling the page file on the D drive so Windows reverts to only using the C drive, like before. It's worth a look.

4. Some engines are more sensitive to HDD performance than others. Unreal Engine does this ALL the time. SSDs have all but eliminated the stutter this causes, but spinny disks really make it noticeable. Games that are 32-bit that use more than 2GB RAM most commonly will repeatedly hit the page file when loading environments or while streaming textures. If you have an option to run the 64-bit version of a game, do it (if you have enough RAM).
 

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
i disagree on point 3

i have been running this benchmark since i built my system and through multiple reformats my scores have always been less than 1% different, as one would expect running a synthetic benchmark. the only thing different now is i have 2 hdds hooked up, i dont understand the drag on performance i am getting. this is going to ruin a lot of my games because my MIN framerates are 300% worse now
 

bystander36

Diamond Member
Apr 1, 2013
5,154
132
106
i disagree on point 3

i have been running this benchmark since i built my system and through multiple reformats my scores have always been less than 1% different, as one would expect running a synthetic benchmark. the only thing different now is i have 2 hdds hooked up, i dont understand the drag on performance i am getting. this is going to ruin a lot of my games because my MIN framerates are 300% worse now

Have you considered that after hooking up this added HD, that you have blocked some air flow? With the way some of the new cards boost, and hoover at a specific temp, perhaps you have just lowered the effective operating clocks of your GPU slightly.
 

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
i didnt consider that mostly because I am monitoring temps and they are not just good, they are great. my system has always ran cool.
 
Feb 25, 2011
16,984
1,616
126
So pull out the new drive and revert the config change, see if that restores your performance.

A 500GB HDD is probably a few years old - could be starting to throw errors and the stuttering is just error correction and read retries, remapping bad sectors, etc.
 

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
I noticed the difference immediately in Wolfenstein TNO which iirc has an engine that uses tons of streaming. My average fps is nearly identical, its the minimum fps drops that are killing the experience, they are much worse now.
 
Last edited:

Flapdrol1337

Golden Member
May 21, 2014
1,677
93
91
After unigine running one loop everything should be in ram, and disks shouldn't matter.

Although, being a benchmark you'd suspect it to load everything before running, in which case it shouldn't matter in the first loop either.
 

chimaxi83

Diamond Member
May 18, 2003
5,457
63
101
You're expecting the same exact score from a benchmark, each time you run it? Do you have any clue what you're doing?

BRB, I'm sending a strongly worded letter to AMD regarding their new driver. I used to get 6809 3DMarks, now I only get 6804. I want a refund.
 

futurefields

Diamond Member
Jun 2, 2012
6,470
32
91
You're expecting the same exact score from a benchmark, each time you run it? Do you have any clue what you're doing?

BRB, I'm sending a strongly worded letter to AMD regarding their new driver. I used to get 6809 3DMarks, now I only get 6804. I want a refund.

You're reaching here.

Read the thread more carefully, each of my posts.
 

bystander36

Diamond Member
Apr 1, 2013
5,154
132
106
It sounds like you may have another issue if you are seeing wide spread problems. A bad HD might cause that, but mostly HD's have little to do with performance, and everything to do with load times.

Perhaps you knocked your CPU HSF loose when you installed the HDD, causing temp issues on the CPU. Perhaps something else is going on. HDD should have no effect on a benchmark.