• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Unigine Heaven 4 benchmark

Soldato
Joined
8 Dec 2004
Posts
14,934
Location
Hampshire
Try this curve (below)
76c is fine, don't worry. It will throttle down if it starts getting too hot. I've had mine up to 87c.
Warranty covers overclocking and even over volting and volt mods on the hardware. Short of taking a blowtorch to your card, you'll be covered

uaIZ2EI.jpg

Cheers Guest2. What do you mean by throttle down? Do you mean downclock? Because this is something I have noticed if I have my case fans on vs off.
And also if I manually whack the GPU fan to 100% it does make a difference to the score it seems since the GPU is downclocking on the fly as the GPU gets hotter it seems...
Ive only just noticed this now!
Regarding the graph above, any reason why this particular profile? Doesnt seem overally different to the default?
Also how does one access the default GPU profile? Presumably MSI wont show this and its somehow programmed into the bios?


i'd say change the curve for benchmarking purposes, the cooler the hardware the better for sustained higher boost clocks, don't worry about it being loud.
For everyday stuff and gaming the stock fan curve will be plenty, and yes msi afterburner is by far the best gpu utility for overclocking and setting up custom fan profiles :)

i run full custom watercooling in my rig and have noctua nf-a12 fans which are really quite at any rpm i set upto 1600rpm, i set 1600 as thats the best spot I've found for cooling vs noise under heavy load senario's, otherwise in games and windows between 1000-1250rpm is the norm

Cheers wookiee87. Yes Im beginning to see that having your rig on max cooling yields best results in Heaven, but I dont see any difference in 3dmark weirdly enough.

I need to look at my the fans on my GPU via the Corsair H100i later.

Scoreboard updated.:)

Cheers for this Kaapstad!

had to do a 1440p run :)

3900x @4.4ghz, 2080ti @2100mhz core 8150mhz memory

JVtt0i1.jpg.png

Im up for that :)
What res is 1440p exactly? 3440 x 1440? I cant see this selection....
 
Soldato
Joined
28 Aug 2017
Posts
2,755
Location
United Kingdom
Cheers Guest2. What do you mean by throttle down? Do you mean downclock? Because this is something I have noticed if I have my case fans on vs off.
And also if I manually whack the GPU fan to 100% it does make a difference to the score it seems since the GPU is downclocking on the fly as the GPU gets hotter it seems...
Ive only just noticed this now!
Regarding the graph above, any reason why this particular profile? Doesnt seem overally different to the default?
Also how does one access the default GPU profile? Presumably MSI wont show this and its somehow programmed into the bios?




Cheers wookiee87. Yes Im beginning to see that having your rig on max cooling yields best results in Heaven, but I dont see any difference in 3dmark weirdly enough.

I need to look at my the fans on my GPU via the Corsair H100i later.



Cheers for this Kaapstad!



Im up for that :)
What res is 1440p exactly? 3440 x 1440? I cant see this selection....

2560 x 1440, if your screen supports it :)

3dmark is a combination of factors tbh, its very hard to increase scores if you post a high one to begin with, when my new ripped cl14 kits arrive I'm going to take one final shot at 1080p and 1440p heaven and 3dmark, the low latency of the ram will help no end, i have cl18 atm and it's utter poo, cant drop timings or bump up speed without pc not posting :(
 
Soldato
Joined
6 May 2009
Posts
19,885
Cheers Guest2. What do you mean by throttle down? Do you mean downclock? Because this is something I have noticed if I have my case fans on vs off.
And also if I manually whack the GPU fan to 100% it does make a difference to the score it seems since the GPU is downclocking on the fly as the GPU gets hotter it seems...
Ive only just noticed this now!
Regarding the graph above, any reason why this particular profile? Doesnt seem overally different to the default?
Also how does one access the default GPU profile? Presumably MSI wont show this and its somehow programmed into the bios?

Yep, it means the GPU will automatically downclock if it gets too hot. Sometimes this will cause it to drop way under normal load clocks, which results in the ~10fps Heaven benchmark minimum fps score.
The graph above is just a custom one i found worked best with my 1080Ti and kept it the same. To use the default one, you just turn off 'enable user defined software automatic fan control'

With my card, the steppings are in 15Mhz increments. I expect all 2080Ti's are the same. With +165Mhz on core I see the following

<60c - 2070Mhz
<65c - 2055Mhz
<70c - 2040Mhz
<75c - 2025Mhz
<80c - 2010Mhz
<85c - 1995Mhz

Try Heaven 4 on windowed mode and watch the results, like so. Note the Max 2070Mhz when it was 48c, then gradually dropping down in 15Mhz stages.

It will not be like this for every card, some will be better than others, some will be absolute donkeys. I've had both types over the years. I think the card I have now is about average in terms of overclock ability. The more expensive card like FTW3 and Kingpin have better cooling - 3 fans, water, or hybrid and Kingpin EVGA card they save/find the ones with good memory and core chip that clocks well and charge more money for them.

Interestinlgy when monitoring you will see things like drop in GPU usage to 73%, this resulted in the fan speed dropping from 100% to 98% which then had the knock on effect of the temp raising to 81c and core clock dropping to 1995Mhz. Notice the tiny blips on the line line

KxWa81X.jpg.png
 
Soldato
Joined
8 Dec 2004
Posts
14,934
Location
Hampshire
2560 x 1440, if your screen supports it :)

3dmark is a combination of factors tbh, its very hard to increase scores if you post a high one to begin with, when my new ripped cl14 kits arrive I'm going to take one final shot at 1080p and 1440p heaven and 3dmark, the low latency of the ram will help no end, i have cl18 atm and it's utter poo, cant drop timings or bump up speed without pc not posting :(

Um... I forget about RAM timings etc, with CL the lower the better right?

I got this stuff coming in tomorrow for me: https://www.corsair.com/uk/en/Categ...emory-Kit/p/CML32GX3M4A1600C10#tab-tech-specs

Any good?

(DDR3 is necessary since I am using an LGA1150 board).


Yep, it means the GPU will automatically downclock if it gets too hot. Sometimes this will cause it to drop way under normal load clocks, which results in the ~10fps Heaven benchmark minimum fps score.
The graph above is just a custom one i found worked best with my 1080Ti and kept it the same. To use the default one, you just turn off 'enable user defined software automatic fan control'

With my card, the steppings are in 15Mhz increments. I expect all 2080Ti's are the same. With +165Mhz on core I see the following

<60c - 2070Mhz
<65c - 2055Mhz
<70c - 2040Mhz
<75c - 2025Mhz
<80c - 2010Mhz
<85c - 1995Mhz

Try Heaven 4 on windowed mode and watch the results, like so. Note the Max 2070Mhz when it was 48c, then gradually dropping down in 15Mhz stages.

It will not be like this for every card, some will be better than others, some will be absolute donkeys. I've had both types over the years. I think the card I have now is about average in terms of overclock ability. The more expensive card like FTW3 and Kingpin have better cooling - 3 fans, water, or hybrid and Kingpin EVGA card they save/find the ones with good memory and core chip that clocks well and charge more money for them.

Interestinlgy when monitoring you will see things like drop in GPU usage to 73%, this resulted in the fan speed dropping from 100% to 98% which then had the knock on effect of the temp raising to 81c and core clock dropping to 1995Mhz. Notice the tiny blips on the line line

KxWa81X.jpg.png

Cheers Guest2. All above is logical and fully understandable.
The fan seems to be going at a bit more pelt using the graph you provided, ergo more noise (but not bothersome actually - the fan on this trio is quite good I have to say!).
Hmm may revert back to the original one... swings and roundabouts.
 
Soldato
Joined
28 Aug 2017
Posts
2,755
Location
United Kingdom
Um... I forget about RAM timings etc, with CL the lower the better right?

I got this stuff coming in tomorrow for me: https://www.corsair.com/uk/en/Categ...emory-Kit/p/CML32GX3M4A1600C10#tab-tech-specs

Any good?

(DDR3 is necessary since I am using an LGA1150 board).


cl9/cl10 ddr3 is pretty much the quickest timing ram you can get, it should give a uplift over your current kit, i've got some cl14 3600mhz ddr4 on the way so going from cl18 that should give a bump in total overall scores, fps should be slightly higher which helps benchmarking and some gaming.

Just another question, do you guys max out "Power Limit (%) and Temp Limit (C)" on Afterburner?

if your cards on air then increase the power limit, but temp limit i'd say leave as is, just for protection running close to 88 degrees is ok short term but with rtx cards the cooler the better, you may end up hurting your score as the card gets too hot and subsequently downclocks because of increased temps.

i max both out but I'm on water so the temp side is pointless as i never see above 50 degrees, my card will pull 133% somehow even though the slider maxe's out at 112%, don't know why but I'm not complaining XD
 
Soldato
Joined
8 Dec 2004
Posts
14,934
Location
Hampshire
Sweet cheers wookie87!

How do you know if the sticks (which I mentioned above) are CL9 or 10?
In one part of the spec it says 9-9-9-24 and another it says 10-10-10-27.

I guess the first numbers are achieved by enabling XMP?
 
Soldato
Joined
28 Aug 2017
Posts
2,755
Location
United Kingdom
Sweet cheers wookie87!

How do you know if the sticks (which I mentioned above) are CL9 or 10?
In one part of the spec it says 9-9-9-24 and another it says 10-10-10-27.

I guess the first numbers are achieved by enabling XMP?

under the tech specs on the link the ram can run 1600mhz @10/10/10/27 (1.5v) or 1333mhz @9/9/9/24 (1.5v) you'll have to have a play and find which settings work best and give the best score :)

enabling xmp should put the kit into 1600mhz at cl10 timings which is still pretty quick :)
 
Soldato
Joined
28 Aug 2017
Posts
2,755
Location
United Kingdom
Question is whats better, tighter timings or faster frequency....

ideally best speed and lowest timings, but it can be difficult in achieving this, and the slightest change can affect you overall score or performance, with your kit you won't notice anything tbh the difference between 9 and 10 is so marginal that i'd set 1600mhz at cl10, (ram speed will help) and call it a day :)
 
Soldato
Joined
6 May 2009
Posts
19,885
ideally best speed and lowest timings, but it can be difficult in achieving this, and the slightest change can affect you overall score or performance, with your kit you won't notice anything tbh the difference between 9 and 10 is so marginal that i'd set 1600mhz at cl10, (ram speed will help) and call it a day :)

What do RAM timings look like now? This is RAM from ~8 years ago

SkSKnCr.jpg.png
 
Soldato
Joined
28 Aug 2017
Posts
2,755
Location
United Kingdom
What do RAM timings look like now? This is RAM from ~8 years ago

ddr4 these days can be anywhere from cl14 up to cl18 and as ever its always best to try and get the fastest timing ram you can (as long as your wallet allows) , downside the quicker and faster you go the more you pay, to the point you pay a couple of hundred pounds for a 100 point uplift in benchmarks, at that point its daft.

I'm running a cl18 kit atm :(, but i have some cl14 inbound, ryzen's have always loved high speed and low timing ram, and now the new 8pack ripped have arrived i thought what the hey and ordered a couple of kits :)

ueEmVeE.png
 
Soldato
Joined
6 May 2009
Posts
19,885
I see the max bandwidth and DRAM frequency is much higher on that RAM, but the CAS timings are also much higher. Arent lower CAS timings better than higher?
It has been years but remember farting around for some time to get it stable at 9-9-9-27 mainly with the 1.5 V. Quite sure the default timings were not great out the box
 
Soldato
Joined
28 Aug 2017
Posts
2,755
Location
United Kingdom
I see the max bandwidth and DRAM frequency is much higher on that RAM, but the CAS timings are also much higher. Arent lower CAS timings better than higher?
It has been years but remember farting around for some time to get it stable at 9-9-9-27 mainly with the 1.5 V. Quite sure the default timings were not great out the box

my cl18 kit is pretty bad tbf, that's why i have a new much faster set arriving, 3600mhz @ 14/15/15/35 coupled with ryzen it should scream through benchmarks hopefully :rolleyes:, no matter what i do with my current kit of ram, my pc will not boot if i change the timings in any direction :( could be down to the manufacture hynix, they do make some awful stuff, not always but on my occasion they did
 
Soldato
Joined
8 Dec 2004
Posts
14,934
Location
Hampshire
Cheers wookie87 and Guest2.
My sticks of RAM will be here later.
Will give Star Citizen a crack since I heard its memory intensive?

(I got these really in preparation for Flight Simulator - I heard thats memory intensive too).
 
Soldato
Joined
6 May 2009
Posts
19,885
my cl18 kit is pretty bad tbf, that's why i have a new much faster set arriving, 3600mhz @ 14/15/15/35 coupled with ryzen it should scream through benchmarks hopefully :rolleyes:, no matter what i do with my current kit of ram, my pc will not boot if i change the timings in any direction :( could be down to the manufacture hynix, they do make some awful stuff, not always but on my occasion they did

Areee, Hynix. I had some Hynix RAM in the past (I think in Gskill too) and it was garbage. Well, i say garbage. It worked but changing anything on it at all would cause crashes or my machine not to post at all
 
Soldato
Joined
6 May 2009
Posts
19,885
Thought I would have a run at 3440x1440 seems as that's the res I play at

1 GPU, 3440x1440
Score 2522, GPU 2080 Ti @2040/4000, CPU 4790k @4.5, Guest2

That's a 43% increase on my 1080Ti (30fps increase from 70.1)

M26WmC2.jpg.png
 
Soldato
Joined
28 Aug 2017
Posts
2,755
Location
United Kingdom
well got my cl14 kit installed and ran again, very nice boost to my original score :D

3900x @4.4ghz all core 2080ti, @2085 core, @8000mhz ram


RXKkuRD.png

VqjmAaf.png

I'm almost at your heel's @Guest2, my gpu has a tiny bit more to give too :), i cant believe just switching from cl18 to cl14 could increase the score that much
 
Last edited:
Back
Top Bottom