is my ram bottle-necking my gpu?

elloguvs

Reputable
Jul 18, 2015
17
0
4,520
hi guys so recently i bought a gigabyte gtx 980ti and its been a bit crap (on a 1080p monitor, a 1440p monitor is on its way)

there has been a small increase in performance since my previous (strix gtx 970) gpu.

i have checked all the thermals for gpu and cpu but all is fine so its not thermal throttling.

i'm thinking its the ram (8gb@1333mhz) that's bottle-necking it but i have no way i can think of to test it.

my specs are:
i7 2600
8gb ram @ 1333mhz
gigabyte gtx980ti
ocz 1250 watt gold psu (yes i know this is overkill for my rig)

will really appreciate any help.
cheers.
 
Solution
Well the reason for running tests is to make comparisons when...

a) Over time to see if things are changing ? degrading over time, and
b) See what the impact is when changing hardware.
c) Use the monitoring software to see if you are throttling

As time hasn't lapsed between tests and you haven't changed hardware, the value of running these tests for you is to check for thermal throttling. You didn't accomplish that w/ this test... read the rest of Item 2 again

1. Run ROG RB and record core voltages and temps with HWiNFO ... since you don't have the CPU overclocked, the CPU should be fine but worth checking in case the case ventilation is inadequate.

2.a Run Furmark as described (till curve flattens .. usually about 20...

misteriosly

Reputable
Jun 1, 2015
605
1
5,360
Probably its your CPU, what games are you playing ?
Download MSI Overburner and monitor how your CPU TEMP/Usage is, also in the same time you could monitor GPU and RAM usage too.
That would help you define what the bottle neck is.
Whats your HDD ?
 

With an intel system Ram speed is not as important as with a Ryzen build. I think it is your 2600 that is the bottle neck. If it was a 'K' chip I'd say overclock it.

 

elloguvs

Reputable
Jul 18, 2015
17
0
4,520


cpu usage during gaming is normally running around the 60%-80% capacity, and temps never go over 65°, could that still be the cpu?

not sure about the specs of my hdd but its a wd 3tb and for my os and some of my most played games are on a "Western Digital Blue PC SSD 250GB"

i play games like bf1 metro redux gta v tomb raider witcher series, things along those lines. (so to my knowledge gpu heavy game)
 
Your GPU is bottlenecking your CPU at 1080p.
Wait until you get a 1440p and you should see better performance if you compare the 980ti vs 970.
1080p is not a hard load on a CPU like the 980ti so its in a way saying " FEED ME " and your cpu tryes hard to do it. As resolution goes up the GPU has to do more calculations = the cpu dont have to transmit so mutch data.
 
The 2600 is a fine CPU and remains a perfectly viable gaming platform. While GPU performance gains have been as much as 50+% faster then the previous generation, CPU gains generation to generation have been single digits, never exceeding 5%. We have a 2600k here in the building w/ a 1080 and multiple monitors and it runs just fine, tho it's an OC'd 2600k.

As far as RAM goes, I could argue both sides of the argument and "prove it".

I could argue that RAM amount makes no difference above 8 GB and then "prove it by showing you several games that show no performance difference whatsoever. I could argue that RAM amount makes a difference above 8 GB and then "prove it by showing you several games that show significant performance difference going from 8 to 16GB.

I could argue that RAM speed makes no difference above 1600 and then "prove it by showing you several games that show no performance difference whatsoever. I could argue that RAM makes a difference above 1600 and then "prove it by showing you several games that show significant performance differences going from 1600 to 2400. Here we see Metro gaining nothing from faster RAM ... and in the next comparison, F1 gains 11%.

http://www.tomshardware.com/reviews/32-gb-ddr3-ram,3790-10.html

1. I would love this to be a 2600k as it's a very overclockable CPU.

2. Before spending money, I would do 2 things:

a. Download RoG Real Bench, Furmark and HWiNFO... run RoG RB for the 8 minute benchmark test, record max core voltages and temps from HWiNFO. Run Furmark till the temp file flattens, record GPU temps.

b. Install MSI Afterburner and try these settings

http://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_980_ti_g1_gaming_soc_review,35.html

With AfterBurner we applied the following settings:

Temp Target 85 Degrees C
CPU clock +150 MHz (from default 1152 MHz)
Mem clock +400 MHz (double datarate is 2x 400 MHz effective)
Voltage + 50Mv
FAN RPM 55% (recommended but a little more is noisy)

Should give you

Core Clock 1302 MHz
Boost Clock: 1452~1477MHz
Memory Clock: 7818 MHz

If it gets up near 82C, I'd increase the fan speed.

3. I am assuming that you didn't wipe drivers in between swapping out the old and new cards.

a. Uninstall all things nVidia from Control Panel - Add / Remove programs
b. Use a registry cleaner (i.e. Ccleaner) to remove all references to nVidia ... and NOTHING else. There are about 15 boxes on the left hand side... check the 1st one and hit "Scan for Issues". Uncheck all the boxes and then check the boxes for any line w/ the word "nVidia" in it. Then select "Fix selected issues'. Now uncheck the 1st box and check the second and repeat. "Rinse and repeat" for all boxes in left panel till done and reboot.
c. Reinstalll all nVidia drivers and programs.

4. If that brings a solution, you can stop there. If ya still want to try the
Without knowing ya MoBo, it's hard to make component recommendations. But this is what we are putting in new builds.... not specifically those brands and models but the speed / CAS is at the point just before getting faster really starts to increase proces ... they were just the cheapest on PCPP today at those specs.... and I prefer low profile RAM.

DDR3 ($118) - https://pcpartpicker.com/product/wxzv6h/gskill-memory-f32400c10d16gtx
DDR4 ($136) - https://pcpartpicker.com/product/sykwrH/gskill-memory-f43200c16d16gtzb

5. The 1st would fit your MoBo assuming the RAM slot layout allowed no interference between the slot layout and an aftermarket cooler. If you don't have a k series CPU, you wouldn't need one.

6. One way to detect whether CPU is bottlenecking is that there will be large performance gap between playing the same game in single versus multiplayer. There is always a gap, but if it is surprisingly large, then the CPU should be on ta list of "probable causes".
 

elloguvs

Reputable
Jul 18, 2015
17
0
4,520


holy crap, thanks for the answer, i'll be trying most of these and hopefully get back to you with some results.
again, thanks for the bloody detailed answer.
 

elloguvs

Reputable
Jul 18, 2015
17
0
4,520


so i ran the rog benchmark and these are the scores:
Image Editing
131,569
Time:40.4957


Encoding
46,428
Time:114.757


OpenCL
96,473
KSamples/sec: 17739


Heavy Multitasking
55,877
Time:136.584


System Score
82,586

Max GPU Temp: 72°C

no clue if thats good or not but i'm hoping you'd know...


also my furmark benchmark results
SCORE:6600 points (110 FPS, 60000 ms)

Resolution: 1920x1080 (FS) - AA:0 samples

FPS: min:94, max:113, avg:110 - OPTIONS: DynBkg

furmark stress test: (1080p 8xmsaa) had an average fps of 34, temps leveled out at 68-70, power% was stable throughout at 82%. didn't overclock anything yet by the way.

oh and weirdly when i stressed it at 2160p 8xmsaa it had the same results. not sure if thats because of my 1080p monitor or...

sorry no screenshots, can't figure out a way to upload them
 
Well the reason for running tests is to make comparisons when...

a) Over time to see if things are changing ? degrading over time, and
b) See what the impact is when changing hardware.
c) Use the monitoring software to see if you are throttling

As time hasn't lapsed between tests and you haven't changed hardware, the value of running these tests for you is to check for thermal throttling. You didn't accomplish that w/ this test... read the rest of Item 2 again

1. Run ROG RB and record core voltages and temps with HWiNFO ... since you don't have the CPU overclocked, the CPU should be fine but worth checking in case the case ventilation is inadequate.

2.a Run Furmark as described (till curve flattens .. usually about 20 minutes) at your current settings and record temps

2.b Run Furmark as described (till curve flattens .. usually about 20 minutes) at your the suggested overclocked settings and record temps ... make sure you are < 85C
 
Solution