Slow graphics -- new thread.

[Doug Laidlaw]

To recap, I am running Mga RC on a system intended to be "average;"
neither slow for office only, nor a gaming box:

Motherboard: Gigabyte H170M with 4 GB RAM.
CPU: ‎Intel(R) Core(TM) i5-6400 CPU @ 2.70GHz 4 cores.
Graphics Card: Nvidia ‎GM107 [GeForce GTX 750]

I tested the RAM to the end of the first cycle of Memtest.

I have reduced the number of cores available for Boinc from 3 to 2,
and set it to use 30% of CPU. GPU is not detected in Boinc.

FPS: Teapot says 60 fps unloaded, 45 fps at worst.

A magazine article says that for games, 60 FPS is fine. The minimum FPS
is 25. With anything lower, movement will be jerky.

Yet to get smooth movement in my Windoze jigsaw program, running under
Crossover Office, I still need to disable Boinc. Otherwise, screens load
like window blinds, and dragging the jigsaw pieces is jerky.

It seems to me that on the above specs, everything should run happily
together. I re-seated the CPU cooler, and got tons of dust out.
Current temp is 30 C. I didn't look at the fan on the video card.

What further tests can I apply?

TIA,

Doug.

[Doug Laidlaw]

Post by Doug Laidlaw
I have reduced the number of cores available for Boinc from 3 to 2,
and set it to use 30% of CPU. GPU is not detected in Boinc.

After sending the above I noticed that CPU was running at 100% (measured
with GkrellM reading only the first core.) The culprit was Boinc, using
100% of CPU in top. I disabled Boinc, and CPU usage shown in 'top'
dropped to 0.3-0.4%.

Doug.

[Doug Laidlaw]

Post by Doug Laidlaw

Post by Doug Laidlaw
I have reduced the number of cores available for Boinc from 3 to 2,
and set it to use 30% of CPU. GPU is not detected in Boinc.

After sending the above I noticed that CPU was running at 100% (measured
with GkrellM reading only the first core.) The culprit was Boinc, using
100% of CPU in top. I disabled Boinc, and CPU usage shown in 'top'
dropped to 0.3-0.4%.
Doug.

Now, with Boinc reduced to 2 cores, and set to run only during periods
of inactivity, I have no problem, except that I think that "inactivity"
never occurs.

I heard of ***@home from a German game server while I was in Swan Hill.
It was very new then. I took my computer in for repair, and they put
SETI on every screen in the shop. I have done enough for them over the
years. They will probably never find aliens, but the technology has had
spin-offs in other directions: a mailout mentioned something like an
x-ray program. Boinc is now being used for other research.

I will keep your reply to post results.

Doug.

[Doug Laidlaw]

Post by Doug Laidlaw
Now, with Boinc reduced to 2 cores, and set to run only during periods
of inactivity, I have no problem, except that I think that "inactivity"
never occurs.

[OT - Tech tip]
I meant to mention that an online HOWTO said: Before adding the final
amount of heatsink paste, rub a little bit into each of the two
surfaces. Get a thin film on each. That fills any microscopic pits and
makes for better heat transfer. It must have worked. My CPU
temperature is now only 19 deg C, and before this, it was never below 25.

Doug.

[Bit Twister]

Post by Doug Laidlaw

Post by Doug Laidlaw
Now, with Boinc reduced to 2 cores, and set to run only during periods
of inactivity, I have no problem, except that I think that "inactivity"
never occurs.

[OT - Tech tip]
I meant to mention that an online HOWTO said: Before adding the final
amount of heatsink paste, rub a little bit into each of the two
surfaces. Get a thin film on each. That fills any microscopic pits and
makes for better heat transfer.

Might depend on paste being used.

I have a really old tube of heat sink paste. Instructions indicated
you were to apply paste to sink and cpu chip as thin as possible to
fill small pits.

That allowed for the most metal to metal contact for max heat transfer.

Warning: Applying too much will result in less physical contact and
eventually lead to cpu failure.

[Doug Laidlaw]

Post by Bit Twister
That allowed for the most metal to metal contact for max heat transfer.

I am using a "house brand" product from a chain like Radio Shack, that I
know is reputable.

In hindsight, 90% of my heat problems may have been due to the fins of
the heat sink being full of dust. There was so much there, that heat
extraction must have been compromised.

[Bit Twister]

Post by Doug Laidlaw
FPS: Teapot says 60 fps unloaded, 45 fps at worst.

Just for fun, edit your ~/.bash_profile and add the line
export vblank_mode=0
save the change and exit editor.

Now click up a terminal and run teapot for a few seconds until you
can pick out an average value. Do not change anything in teapot's screen.

Log out/in and re-run the teapot test.

On my ATI Radeon HD 5000 to HD 6300 on cauldron I get 60fps before the
export log out/in. With the export I get ~850fps upon log in.

I find the teapot test fps closer to real world speed with the teapot
menu showing.

Without the menu it shows ~1225fps.

[Doug Laidlaw]

Post by Bit Twister

Post by Doug Laidlaw
FPS: Teapot says 60 fps unloaded, 45 fps at worst.

Just for fun, edit your ~/.bash_profile and add the line
export vblank_mode=0
save the change and exit editor.
Now click up a terminal and run teapot for a few seconds until you
can pick out an average value. Do not change anything in teapot's screen.
Log out/in and re-run the teapot test.
On my ATI Radeon HD 5000 to HD 6300 on cauldron I get 60fps before the
export log out/in. With the export I get ~850fps upon log in.
I find the teapot test fps closer to real world speed with the teapot
menu showing.
Without the menu it shows ~1225fps.

That gave me a steady 60 fps, with a couple at 59.5. Only the very
first count (32.5 fps) was significantly different. I shouldn't need a
complete reboot. Maybe something else is acting as a governor?

I recall reading about that option before. Opinions were divided on the
effect it would have on rendering.

[Bit Twister]

Post by Doug Laidlaw

Post by Bit Twister

Post by Doug Laidlaw
FPS: Teapot says 60 fps unloaded, 45 fps at worst.

Just for fun, edit your ~/.bash_profile and add the line
export vblank_mode=0
save the change and exit editor.
Now click up a terminal and run teapot for a few seconds until you
can pick out an average value. Do not change anything in teapot's screen.
Log out/in and re-run the teapot test.
On my ATI Radeon HD 5000 to HD 6300 on cauldron I get 60fps before the
export log out/in. With the export I get ~850fps upon log in.
I find the teapot test fps closer to real world speed with the teapot
menu showing.
Without the menu it shows ~1225fps.

That gave me a steady 60 fps, with a couple at 59.5. Only the very
first count (32.5 fps) was significantly different. I shouldn't need a
complete reboot. Maybe something else is acting as a governor?
I recall reading about that option before. Opinions were divided on the
effect it would have on rendering.

Well, with release 6 not supporting Radeon propitiatory driver, I have
to set it if I want to play something like GL-117 which will tell you
it will not run with a low FPS.

Even with a teapot 838fps, GL-117 shows about 80fps on my cauldron
install.