-
Hi Guest!
This is a notice regarding recent upgrades to the Hub. Over the last month, we have added several new features to improve your experience.
You can check out the details in our official announcement!
17 FPS on 3060
- Thread starter NaughtyAI
- Start date
- Messages
- 655
- Reactions
- Points
- 63
You have 3 different timings on your RAM with 2 different clock speeds ... oh boy. 2666MHz also is no winner.
What about the most important, GPU-Z --> Graphics card??
What about the most important, GPU-Z --> Graphics card??
Yup. 3DMark tells us there is clearly something wrong, so at least it's not a VaM problem.
Random ideas:
Internet says your card has a single 8pin power connector? Did you forget to connect it? (Even without it might still "run", as it gets power from the PCIe port as well) Check the manual of your power supply which cable should be used for the graphics card. Usually there is a cable reserved just for the GPU as it draws so much power. Other cables, e.g. one intended for old harddisks and DVD drives might not give you enough power, although using the same connectors.
Unlikely, but just in case, as you changed the power supply....maybe forgot the CPU power connector? Same as the GPU, the CPU get's power from the mainboard, but also from that extra power connector. So it might still run, but not at full clock rate. Internet says your CPU should boost to 4.3GHz, but the benchmark screenshot somewhere above only showed it at 4.0GHz? Does your CPU boost on all cores when you run some CPU benchmark?
Do you still have the old graphics card? You could try to put it back in and see whether performance is as expected (for that old card) again. That would rule out problems with CPU, memory, mainboard, etc.
- Messages
- 655
- Reactions
- Points
- 63
I thought the same but 4.3 is just single core boost. And only for a short amount of time.
But the RAM clock speed is probably a handbrake for CPU performance.
Last edited:
Yes, but that much? And RAM has not changed?
- Messages
- 655
- Reactions
- Points
- 63
Yeah, thats one of the riddles here ... but no solution to some of the questions until we have more answers. Like is the card running with full bandwith.
But PCIe 3 vs 4 is only low percentage performance difference from what I know. So the the problem has to be somewhere else.
But PCIe 3 vs 4 is only low percentage performance difference from what I know. So the the problem has to be somewhere else.
- Messages
- 763
- Reactions
- Points
- 93
Summary of what we know about entire config so far:
CPU: I5-10400F
Board: Gigabyte H510M H (likely rev 1.0/1.1, based on old F7 bios version)
GPU: Palit RTX 3060 (12GB) @2560x1440p 59Hz + 2nd display (based on TimeSpy) + maybe Vsync/Adaptive-sync is on (based on VaM results) locking @60fps
RaM: 16GB in dual channel (2x8GB AMD Radeon R7 Perf. Series (R748G2606U2S-U) 2666Mhz, XMP off)
Storage1: SSD (probably Gen3.M2 or SATA?)
Storage2: HDD
PSU: 750W
Default bios settings (based on xmp off, rebar req. off, auto pcie slot config, also gen1 link speed of 2.5GT/s due to gpu idle... should be 16GT/s on gpu load)
If everything else suggested so far is not the case (connection, cables, etc):
I would try to force disable vsync or any adaptive sync (gsync/freesync) ...just to be sure with maxFPS
As extra I would go inside BIOS and select pcie gen4 for gpu (from auto), enable xmp and all re-BAR stuff
Also I would consider updating bios to F17.
CPU: I5-10400F
Board: Gigabyte H510M H (likely rev 1.0/1.1, based on old F7 bios version)
GPU: Palit RTX 3060 (12GB) @2560x1440p 59Hz + 2nd display (based on TimeSpy) + maybe Vsync/Adaptive-sync is on (based on VaM results) locking @60fps
RaM: 16GB in dual channel (2x8GB AMD Radeon R7 Perf. Series (R748G2606U2S-U) 2666Mhz, XMP off)
Storage1: SSD (probably Gen3.M2 or SATA?)
Storage2: HDD
PSU: 750W
Default bios settings (based on xmp off, rebar req. off, auto pcie slot config, also gen1 link speed of 2.5GT/s due to gpu idle... should be 16GT/s on gpu load)
If everything else suggested so far is not the case (connection, cables, etc):
I would try to force disable vsync or any adaptive sync (gsync/freesync) ...just to be sure with maxFPS
open Nvidia Control Panel (not geforce experience):
-3D settings/manage 3D settings/set Vsync off
-Display/set up G-sync/uncheck enable gsync
(it should be the same in case of freesync display)
Run TimeSpy and compare, in case entire run was limited to 60fps and reporting low score because of it.
-3D settings/manage 3D settings/set Vsync off
-Display/set up G-sync/uncheck enable gsync
(it should be the same in case of freesync display)
Run TimeSpy and compare, in case entire run was limited to 60fps and reporting low score because of it.
As extra I would go inside BIOS
On boot <Delete> key to enter and switch to Advanced Mode <F2> (if inside Easy Mode)
goto:
-Tweaker / Extreme Memory Profile(X.M.P.) / Disabled > Profile 1 (even if same speed, timers are fastened with xmp profile)
-Settings / IOPORTS / Initial Display Output / Above 4G Decoding/ Disabled > Enabled
-Settings / IOPORTS / Initial Display Output / Re-size BAR Support / Disabled > Auto
-Not sure where setting to change from Auto to Gen4 is, but this board supports gen4 (gpu only)
On AMD is PCIe Slot Configuration inside Misc. Nevermind, 10th gen doesn't support gen4. It's still full x16 card, so 8GT/s should be enough.
goto:
-Tweaker / Extreme Memory Profile(X.M.P.) / Disabled > Profile 1 (even if same speed, timers are fastened with xmp profile)
-Settings / IOPORTS / Initial Display Output / Above 4G Decoding/ Disabled > Enabled
-Settings / IOPORTS / Initial Display Output / Re-size BAR Support / Disabled > Auto
On AMD is PCIe Slot Configuration inside Misc.
Also I would consider updating bios to F17.
F7 is kinda old and nowadays updating bios is more "fool proof" than it was (with floppy disks and what not).
Unless power goes out while flashing.... Like seriously what are chances of that to happen?
Seriously...
If you never flash to most recent bios (that released like few days ago) or beta bios (unless you really...REALLY have to)
and don't use flashing tools inside OS (always flash inside bios) ...you should be fine.
Flashing bios on laptop is even safer thanks to battery (no risk of power outage).
Unless power goes out while flashing.... Like seriously what are chances of that to happen?
Seriously...
If you never flash to most recent bios (that released like few days ago) or beta bios (unless you really...REALLY have to)
and don't use flashing tools inside OS (always flash inside bios) ...you should be fine.
Flashing bios on laptop is even safer thanks to battery (no risk of power outage).
Last edited:
As for the additional power connectors, I rebuilt it 3 times, I connected the power for the CPU and GPU 100% correctly
on the CPU, all cores are overclocked to 4.0 GHz, at least that's what the afterburner above says
I'll try changing the video card.
У вас есть 3 разных тайминга в вашей оперативной памяти с 2 разными тактовыми частотами ... о боже. 2666 МГц тоже не победитель.
А что насчет самого важного, GPU-Z --> Видеокарта??
[/ЦИТИРОВАТЬ]
Changed only the power supply and video card
I tried to turn off the G-Sync/Vsync, but the result did not change, the fps remained the same as it was 30
I set up the bios, here is the result. I didn't find the from Auto to Gen4 setting
Last edited:
- Messages
- 763
- Reactions
- Points
- 93
Forget about it, small overlook on my part (10th gen supports only Gen3).
You don't loose much (0-2 fps), it's still full x16 card. Once you load/use gpu, current link speed should go up to 8GT/s (x16 gen3).
Still, 3060 results are way below average when compared to 1050 Ti results.
- Messages
- 655
- Reactions
- Points
- 63
The low average, in my opinion, comes from low RAM speed plus slow CPU. Its no GPU only test.
And compared to 1050ti result everything is fine, around triple performance.
I see no problem here. The problem is the CPU plus RAM and wrong performance measurements in VaM from the user. My thoughts.
And compared to 1050ti result everything is fine, around triple performance.
I see no problem here. The problem is the CPU plus RAM and wrong performance measurements in VaM from the user. My thoughts.
The "average" is skewed because of crazy overclockers, so you can ignore that. What you care about is the orange line.
With the 1050TI you are right on the peak of users, that's where you expect to be:
With the RTX3060 there is clearly something wrong, as you are way below the peak:
Does the RTX3060 look like on the website?
The performance difference could be about right if it were a mobile version of the card. However, it has 12GB (at least is says so) and Palit does not make mobile versions of the RTX3060?
That however is also true. Performance in VaM should be the same (if limited by CPU) or a bit faster, not slower.
You could try the BIOS update as the one from April specifically fixes an RTX3060 issue. (However, not this issue?)
It say: "Address blank display while system powers up with RTX3060 series graphics cards"
But other than that, I'm out of ideas.
- Messages
- 655
- Reactions
- Points
- 63
We are speculating on the VaM performance based on the users observation, that could be simply wrong. Though it makes no sense that results with a 3060 are worse than with a 1050ti, it could be CPU limited anyway.
I mean we dont have 2 benchmark runs in VaM to compare. That would be a basis. But like this ... poking around in the fog.
I mean we dont have 2 benchmark runs in VaM to compare. That would be a basis. But like this ... poking around in the fog.
Lytanshade
Member
- Messages
- 46
- Reactions
- Points
- 18
were you using the 1366x768 display with the 1050ti and then upgraded to the 1440p monitor with the 3060?
- Messages
- 763
- Reactions
- Points
- 93
Honestly, If OP went from 768p to 1440p, that would be nice "playing dumb" move.
True, we don't have 2 VaM benchmark results to compare both GPUs, so everything is based on what OP said.
Yes, 1440p is more tanky (than say 1080p)... still even if it's only 10400F @1333 it should be good enough for other games.
Based on both timespy results @1440p it should perform better or same, not worse.
But somehow it's also worse in other games... Like how?! ...graphics score is 3x (what about old saying... "try to invest more into GPU"?).
Completely unrelated (out of curiosity): I did timespy ram comparison with 3200(1600) and 1333(667)
True, we don't have 2 VaM benchmark results to compare both GPUs, so everything is based on what OP said.
Yes, 1440p is more tanky (than say 1080p)... still even if it's only 10400F @1333 it should be good enough for other games.
Based on both timespy results @1440p it should perform better or same, not worse.
But somehow it's also worse in other games... Like how?! ...graphics score is 3x
(what about old saying... "try to invest more into GPU"?).Completely unrelated (out of curiosity): I did timespy ram comparison with 3200(1600) and 1333(667)
There is noticeable CPU impact @667, but graphics result is about the same.
@1600 XMP (16-18-18-36)
@667 (10-10-10-22)
This is more unrealistic example, in theory if OP had 2933 or 3200 sticks, it wouldn't make a dent (vs 2666).
But any 10600/11600 could (higher all-core and peak boost, +2 cores). I would consider checking used market for 1x600/1x700 parts.
@1600 XMP (16-18-18-36)
@667 (10-10-10-22)
This is more unrealistic example, in theory if OP had 2933 or 3200 sticks, it wouldn't make a dent (vs 2666).
But any 10600/11600 could (higher all-core and peak boost, +2 cores). I would consider checking used market for 1x600/1x700 parts.
- Messages
- 655
- Reactions
- Points
- 63
One thing is for sure: MacGruber has not retired yet 
cbkstudios
Active member
Check this...
Control Panel / Hardware and Sound / Power Options
You should see a list of power plans there:
You can go to Change Plan Settings then Advanced Settings. You're looking for the GPU and CPU to be set to Maximum Performance.
