SUBSIM Radio Room Forums



SUBSIM: The Web's #1 resource for all submarine & naval simulations since 1997

Go Back   SUBSIM Radio Room Forums > General > General Topics > PC Hardware/Software forum
Forget password? Reset here

Reply
 
Thread Tools Display Modes
Old 05-19-18, 01:49 AM   #1
BarracudaUAK
Captain
 
Join Date: Apr 2016
Posts: 520
Downloads: 31
Uploads: 0


Default New Video card - Benchmarks!

As I mentioned in the thread about the monitor, I found a video card that was -moderately- affordable. (Not really, but it was better than they were.)

Also as I posted here:
http://www.subsim.com/radioroom/showthread.php?t=232683
in the thread about new PC options, in March of this year the RX580 was ~$500.

Found a XFX RX 580 for ~$370, so I decided to get it now since the last time I waited they were all out of stock for months, AND doubled in price.

Had to go to 2 locations to find 1 (the last one), but now I have it!
All I need now is the rest of the over-priced parts and I'll have a new PC!

Short Version: Only in benchmarks that put most of the load on the GPU did the newer card really make a difference.
I've read a few places that the RX 580 and RX Vega at 1920x1080 are "CPU Limited".


EDIT: Or maybe not? I finally found the console command to increase the FPS cap on Source Engine games... added below.


ON TO THE BENCH MARKS!

First the stuff that didn't change:


AMD FX8350 8-core 4.0ghz (4.2 turbo, does this actually work?)
Gigabyte GA-990FXA-UD3 (insert sarcasm here.)
32GB Corsair DDR3 (I forget the specs)
OS: Fedora 28 (Linux, not a hat )


The 2 video cards are:


Gigabyte Radeon R9 380 2GB

Link to specs: https://www.gigabyte.com/Graphics-Ca...D-rev-10-11#sp


XFX Radeon RX 580 8GB
Link to specs: http://www.xfxforce.com/en-us/produc...d-rx-580p8dbdr


I'll post the #s by the benchmark.
All benchmarks were run with the CPU governor set to "Performance".
Based on what I've read, Performance runs at the max stock clock speed, any "throttling" is done by the hardware.
But, I could be wrong. Either way, "Performance" sounds good.

-----
Quote:
GLmark2:

"glmark2" runs a series of OpenGL rendering test/benchmarks in a window, appears to be 640x480, and displays in the terminal the frame-rate of each part of the test/benchmark.


R9 380:
glmark2 Score: 7516

Details:
 

[build] use-vbo=false: FPS: 5718 FrameTime: 0.175 ms
[build] use-vbo=true: FPS: 9673 FrameTime: 0.103 ms
[texture] texture-filter=nearest: FPS: 9450 FrameTime: 0.106 ms
[texture] texture-filter=linear: FPS: 9483 FrameTime: 0.105 ms
[texture] texture-filter=mipmap: FPS: 9552 FrameTime: 0.105 ms
[shading] shading=gouraud: FPS: 9628 FrameTime: 0.104 ms
[shading] shading=blinn-phong-inf: FPS: 9666 FrameTime: 0.103 ms
[shading] shading=phong: FPS: 9680 FrameTime: 0.103 ms
[shading] shading=cel: FPS: 9564 FrameTime: 0.105 ms
[bump] bump-render=high-poly: FPS: 8151 FrameTime: 0.123 ms
[bump] bump-render=normals: FPS: 9490 FrameTime: 0.105 ms
[bump] bump-render=height: FPS: 9392 FrameTime: 0.106 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 9826 FrameTime: 0.102 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 7312 FrameTime: 0.137 ms
[pulsar] light=false:quads=5:texture=false: FPS: 8522 FrameTime: 0.117 ms
[desktop] blur-radius=5:effect=blur:passes=1:separable=true:windo ws=4: FPS: 4729 FrameTime: 0.211 ms
[desktop] effect=shadow:windows=4: FPS: 5139 FrameTime: 0.195 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 1115 FrameTime: 0.897 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 1325 FrameTime: 0.755 ms
[buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 1331 FrameTime: 0.751 ms
[ideas] speed=duration: FPS: 3229 FrameTime: 0.310 ms
[jellyfish] <default>: FPS: 7987 FrameTime: 0.125 ms
[terrain] <default>: FPS: 1229 FrameTime: 0.814 ms
[shadow] <default>: FPS: 7541 FrameTime: 0.133 ms
[refract] <default>: FPS: 2324 FrameTime: 0.430 ms
[conditionals] fragment-steps=0:vertex-steps=0: FPS: 9423 FrameTime: 0.106 ms
[conditionals] fragment-steps=5:vertex-steps=0: FPS: 9635 FrameTime: 0.104 ms
[conditionals] fragment-steps=0:vertex-steps=5: FPS: 9616 FrameTime: 0.104 ms
[function] fragment-complexity=low:fragment-steps=5: FPS: 9612 FrameTime: 0.104 ms
[function] fragment-complexity=medium:fragment-steps=5: FPS: 9549 FrameTime: 0.105 ms
[loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 9692 FrameTime: 0.103 ms
[loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 9733 FrameTime: 0.103 ms
[loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 9735 FrameTime: 0.103 ms
================================================== =====
glmark2 Score: 7516
================================================== =====



RX 580:
glmark2 Score: 7717

Details:
 

[build] use-vbo=false: FPS: 5394 FrameTime: 0.185 ms
[build] use-vbo=true: FPS: 9853 FrameTime: 0.101 ms
[texture] texture-filter=nearest: FPS: 9616 FrameTime: 0.104 ms
[texture] texture-filter=linear: FPS: 9565 FrameTime: 0.105 ms
[texture] texture-filter=mipmap: FPS: 9638 FrameTime: 0.104 ms
[shading] shading=gouraud: FPS: 9786 FrameTime: 0.102 ms
[shading] shading=blinn-phong-inf: FPS: 9553 FrameTime: 0.105 ms
[shading] shading=phong: FPS: 9575 FrameTime: 0.104 ms
[shading] shading=cel: FPS: 9591 FrameTime: 0.104 ms
[bump] bump-render=high-poly: FPS: 9590 FrameTime: 0.104 ms
[bump] bump-render=normals: FPS: 9379 FrameTime: 0.107 ms
[bump] bump-render=height: FPS: 9301 FrameTime: 0.108 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 9820 FrameTime: 0.102 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 9817 FrameTime: 0.102 ms
[pulsar] light=false:quads=5:texture=false: FPS: 8709 FrameTime: 0.115 ms
[desktop] blur-radius=5:effect=blur:passes=1:separable=true:windo ws=4: FPS: 5263 FrameTime: 0.190 ms
[desktop] effect=shadow:windows=4: FPS: 5286 FrameTime: 0.189 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 1038 FrameTime: 0.963 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 1338 FrameTime: 0.747 ms
[buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 1211 FrameTime: 0.826 ms
[ideas] speed=duration: FPS: 3265 FrameTime: 0.306 ms
[jellyfish] <default>: FPS: 8005 FrameTime: 0.125 ms
[terrain] <default>: FPS: 1845 FrameTime: 0.542 ms
[shadow] <default>: FPS: 7597 FrameTime: 0.132 ms
[refract] <default>: FPS: 3389 FrameTime: 0.295 ms
[conditionals] fragment-steps=0:vertex-steps=0: FPS: 9694 FrameTime: 0.103 ms
[conditionals] fragment-steps=5:vertex-steps=0: FPS: 9679 FrameTime: 0.103 ms
[conditionals] fragment-steps=0:vertex-steps=5: FPS: 9607 FrameTime: 0.104 ms
[function] fragment-complexity=low:fragment-steps=5: FPS: 9693 FrameTime: 0.103 ms
[function] fragment-complexity=medium:fragment-steps=5: FPS: 9697 FrameTime: 0.103 ms
[loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 9709 FrameTime: 0.103 ms
[loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 9721 FrameTime: 0.103 ms
[loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 9457 FrameTime: 0.106 ms
================================================== =====
glmark2 Score: 7717
================================================== =====
Quote:
"vkmark " is the same idea, but with Vulkan.

R9 380:
vkmark Score: 11057

Details:
 

[vertex] device-local=true: FPS: 12059 FrameTime: 0.083 ms
[vertex] device-local=false: FPS: 6496 FrameTime: 0.154 ms
[texture] anisotropy=0: FPS: 11641 FrameTime: 0.086 ms
[texture] anisotropy=16: FPS: 11619 FrameTime: 0.086 ms
[shading] shading=gouraud: FPS: 11287 FrameTime: 0.089 ms
[shading] shading=blinn-phong-inf: FPS: 11317 FrameTime: 0.088 ms
[shading] shading=phong: FPS: 11228 FrameTime: 0.089 ms
[shading] shading=cel: FPS: 11191 FrameTime: 0.089 ms
[effect2d] kernel=edge: FPS: 12251 FrameTime: 0.082 ms
[effect2d] kernel=blur: FPS: 8522 FrameTime: 0.117 ms
[desktop] <default>: FPS: 9222 FrameTime: 0.108 ms
[cube] <default>: FPS: 13004 FrameTime: 0.077 ms
[clear] <default>: FPS: 13911 FrameTime: 0.072 ms
================================================== =====
vkmark Score: 11057
================================================== =====



RX 580:
vkmark Score: 13870

Details:
 
Quote:
[vertex] device-local=true: FPS: 15256 FrameTime: 0.066 ms
[vertex] device-local=false: FPS: 7239 FrameTime: 0.138 ms
[texture] anisotropy=0: FPS: 14714 FrameTime: 0.068 ms
[texture] anisotropy=16: FPS: 14739 FrameTime: 0.068 ms
[shading] shading=gouraud: FPS: 14206 FrameTime: 0.070 ms
[shading] shading=blinn-phong-inf: FPS: 14235 FrameTime: 0.070 ms
[shading] shading=phong: FPS: 14113 FrameTime: 0.071 ms
[shading] shading=cel: FPS: 14110 FrameTime: 0.071 ms
[effect2d] kernel=edge: FPS: 15184 FrameTime: 0.066 ms
[effect2d] kernel=blur: FPS: 12000 FrameTime: 0.083 ms
[desktop] <default>: FPS: 11436 FrameTime: 0.087 ms
[cube] <default>: FPS: 16043 FrameTime: 0.062 ms
[clear] <default>: FPS: 17037 FrameTime: 0.059 ms
================================================== =====
vkmark Score: 13870
================================================== =====
Quote:
The Vulkan "Smoketest" is actually a bunch of teapots bouncing around the screen.
Not sure of the resolution, might be 1280x1024. Best guess.


It runs the test, and then outputs the average over 5 seconds.
I let it run and then stopped it when it had 9 "outputs". "Outputs"? "Averages"?
Stopped at 9 either way.


R9 380:

5924 presents in 5.00078 seconds (FPS: 1184.62)
5920 presents in 5.00052 seconds (FPS: 1183.88)
5918 presents in 5.00004 seconds (FPS: 1183.59)
5909 presents in 5.00028 seconds (FPS: 1181.73)
5955 presents in 5.00045 seconds (FPS: 1190.89)
5920 presents in 5.00038 seconds (FPS: 1183.91)
5927 presents in 5.00057 seconds (FPS: 1185.26)
5919 presents in 5.00064 seconds (FPS: 1183.65)
5929 presents in 5.00034 seconds (FPS: 1185.72)


RX 580:

7762 presents in 5.00034 seconds (FPS: 1552.29)
7757 presents in 5.00054 seconds (FPS: 1551.23)
7645 presents in 5.00129 seconds (FPS: 1528.6)
7674 presents in 5.00026 seconds (FPS: 1534.72)
7680 presents in 5.00033 seconds (FPS: 1535.9)
7695 presents in 5.00036 seconds (FPS: 1538.89)
7631 presents in 5.00045 seconds (FPS: 1526.06)
7731 presents in 5.00008 seconds (FPS: 1546.17)
7775 presents in 5.00036 seconds (FPS: 1554.89)
Quote:
Counter Strike Source.

CS:S has a built in benchmark, the "stress test". I did 3 runs, as it changes a bit each time.

Both cards test were at:
1920x1080
Full Screen
8x MSAA
16X Anisotropic Filtering

Motion blur: OFF.
High Dynamic Range: NONE
Bloom: OFF
All other settings were at the highest possible.


Average FPS:


R9 380:

290 fps
285 fps
291 fps




RX 580:

287 fps
288 fps
284 fps


EDIT: Setting Max Frame Rate to 500:


361 fps


Setting Max to 999:


390 fps


End Edit.



I also checked the RX 580 with
Motion blur: ON
High Dynamic Range: ON
Bloom: ON

Not much difference!

RX 580:

284 fps
280 fps
281 fps




I also did 3 runs on the old PC for comparison:
This is the oldest system/Video Cards that I have available ATM.
Should give a better idea what to expect if you jump up to a RX series card.

Athlon X2 5600 Dual-Core 2.8ghz
8GB Patriot 800mhz DDR2
OS: Fedora 28
HD 6750 1GB GDDR5
(GPU and RAM clocks, I don't remember atm.)

EDIT: I forgot to set the governor to "Performance" last time, test re-run with governor now set to "Performance".



HD 6750

141 fps
143 fps
143 fps

Quote:
Unigine Benchmarks:

Only did 1 run each, as it rarely changes.

Unigine Heaven:
1920x1080
Full Screen

"Ultra"
8xAA
Tessellation: Extreme
Note about "min", there is a place in the benchmark where the HD light flashes, and it stutters.
Might be the reason. Haven't tested further.

R9 380:

Score: 761
Average FPS: 30.2
Min FPS: 7.6
Max FPS: 64.6




Rx 580:

Score: 1161
Average FPS: 46.1
Min FPS: 8.3
Max FPS: 97.2
Quote:
Unigine Valley:
Valley is the newer benchmark, but seems to be less demanding.

1920x1080
8x AA
Full screen
"Extreme HD" Preset

R9 380:

Score: 1411
Average FPS: 33.7
Min FPS: 18.9
Max FPS: 60.8




RX 580:

Score: 1997
Average FPS: 47.7
Min FPS: 24.0
Max FPS: 88.1
Quote:
Unigine Superposition


The last benchmark is a real whopper.
With the 2GB on the R9, the highest you can go is "1080p Medium".
Any higher and you need more RAM, so I ran both at "1080p Medium".
Then I ran the Rx 580 at Extreme.
When the benchmark ends, the "game/benchmark" closes and it drops back to the menu to give you the results.
It froze there, I haven't tried again.
HOWEVER, I did take a picture of the stats in the corner. So I have FPS, but no "score" for that setting/run.


R9 380:

Score: 4321
Avg FPS: 32.32
Min FPS: 26.15
Max FPS: 41.83
Temp deg. C:
Min: 41

Max: 73


RX 580:

Score: 6722
Avg FPS: 50.28
Min FPS: 39.08
Max FPS: 66.40
Temp deg C:
Min: 46
Max: 74


Preset: 1080p Extreme

RX 580
FPS at system freeze: 14.5
vg FPS: 15.3
Min FPS: 12.2
Max FPS: 18.2


Further details:

Scene: 17/17
Frames: 2683
Surfaces: 2567
Materials: 359
Light Sources: 9
Trianges: 6,010,893

Final thoughts:
Most of the games that I have will never touch the RX 580's limits, leaving the CPU the bottleneck for now.
If I go with a 4k monitor, then the situation might change.

Barracuda


P.S. Please let me know if the Quote boxes make this easier, or harder to understand.

Last edited by BarracudaUAK; 06-07-18 at 06:15 PM. Reason: Formatting is all over the place!
BarracudaUAK is offline   Reply With Quote
Old 05-19-18, 04:16 PM   #2
Sailor Steve
Eternal Patrol
 
Sailor Steve's Avatar
 
Join Date: Nov 2002
Location: High in the mountains of Utah
Posts: 50,369
Downloads: 745
Uploads: 249


Default

Cool stuff.

I finally have the PC I've always wanted, and it doesn't come close to yours.
AMD Ryzen 5 1400 4-Core 3.2 Ghz
8 Ghz RAM
Radeon RX 560

Like you, I don't see myself ever having a game that will tax that. I only mention it at all because I seem to have made one mistake. I've compared Windows 10 and 8.1 in the past, and a couple of the programs I use prefer 8.1, to the point of not working well with 10 at all. I had the option to downgrade to 8.1, but after doing so I discovered that MS seems to have made a deal with both Intel and AMD, so that none of the newest processors work with anything other than Win 10. It's frustrating to realize that I could have had an older Athlon processor that cost less, would still be more than I needed, and would work with 8.1.

My main reason for posting is to make sure everybody knows about that little problem.
__________________
“Never do anything you can't take back.”
—Rocky Russo
Sailor Steve is offline   Reply With Quote
Old 05-19-18, 11:20 PM   #3
BarracudaUAK
Captain
 
Join Date: Apr 2016
Posts: 520
Downloads: 31
Uploads: 0


Default

Quote:
Originally Posted by Sailor Steve View Post
Cool stuff.

I finally have the PC I've always wanted, and it doesn't come close to yours.
AMD Ryzen 5 1400 4-Core 3.2 Ghz
8 Ghz RAM
Radeon RX 560

Like you, I don't see myself ever having a game that will tax that. I only mention it at all because I seem to have made one mistake. I've compared Windows 10 and 8.1 in the past, and a couple of the programs I use prefer 8.1, to the point of not working well with 10 at all. I had the option to downgrade to 8.1, but after doing so I discovered that MS seems to have made a deal with both Intel and AMD, so that none of the newest processors work with anything other than Win 10. It's frustrating to realize that I could have had an older Athlon processor that cost less, would still be more than I needed, and would work with 8.1.

My main reason for posting is to make sure everybody knows about that little problem.

Actually yours is pretty close!
In several ways, better.


EDIT: Short version, your Ryzen is excellent.



I did a quick search for the Ryzen 1400, and grabbed the specs from AMD's website.
architecture

AMD Ryzen™ 5 1400

# of CPU Cores: 4
# of Threads: 8
Base Clock: 3.2GHz
Max Boost Clock: 3.4GHz
CMOS: 14nm
Package: AM4
PCI Express Version: PCIe 3.0 x16


Compared to my FX 8350:

# of CPU Cores : 8
# of Threads: 8
Base Clock: 4.0GHz
Max Boost Clock: 4.2GHz
CMOS: 32nm
Package: AM3+
PCI Express Version: PCIe 2.0 x16


The FX 8350 doesn't "HyperThread", so I only have 8 threads.
The Ryzens, based on all I've read, for the same clock speed have 40% more IPC (Instructions per clock/instructions per cycle).

So if your Ryzen was a FX cpu, it would be roughly equivalent to a 4.48ghz (bases on some quick math, 3.2 X 1.4 = 4.48) in IPC.

Your CPU is less than half the architectural size of mine = Less heat.

PCIe 3.0 is TWICE AS FAST.
(GiB/s = Gigabytes per Second.)
PCIe 2.0 is ~8GiB/sec max. on a x16 slot.
PCIe 3.0 is ~16 GiB/sec max on a x16 slot.

Meaning IF, for example, you were to run Crossfire with 2 cards. Both would be at x8.
Your board could "feed" each card data at 8GiB/sec. You can feed both cards as fast as I can feed 1.


I seriously considered a RX560, but I couldn't find any #s on the same games or benchmarks like Unigine Heaven/Valley/Superposition.

The driver disk that came with my RX 580 doesn't include Drivers for Win 8/8.1.
The box shows supported Operating Systems as: "Linux, Windows 10, and Windows 7".

To me, Windows 8 has always seemed to be like Vista. A modification to a design that people liked, with serious "structural issues". Vista driver situation (compared to XP) was similar to Win 8/8.1 is (compared to Win7).
I don't see Win 8 drivers on the AMD website for the RX Series!

Also, you are using DDR4, Mine uses DDR3.
Faster there too!
Given the current prices, I would most likely have gone with 8-16GB RAM (depending on the Motherboard).


Keep in mind I built my system over 9 months, first gathering core parts on one trip.
Then case, fans, etc. the next time (mini road trips! kinda).
Then finally ran across a sale at a local store got PSU, R7 370X, and a part that I'm forgetting to get running. This took a few months.
A few months later I got the 2nd R7, and a few months after that, got the last 2 HDD for the RAID.

Just before the RX 480's launch I was out of town again, and the store I had bought most the core parts at, had the R9 380 and the R9 Fury on sale.
The R9 380 was ~$150, and the R9 Fury (Fiji core, HBM ram) was ~$200.
I got the 380s, still kinda wish I had gotten the Fury, it was out running the VEGA in games/benchmarks for several months! Might not have needed to get the RX580!!!

I'm sure with some careful, slow deliberate (read: on sale) parts collection, your Ryzen will be a total monster in a short time.



Barracuda

Last edited by BarracudaUAK; 05-23-18 at 12:48 AM.
BarracudaUAK is offline   Reply With Quote
Old 05-21-18, 01:24 PM   #4
BarracudaUAK
Captain
 
Join Date: Apr 2016
Posts: 520
Downloads: 31
Uploads: 0


Default

I did a quick test with Kerbal Space Program.


Although I'm still looking for the option/setting to show the fps (I know I saw it in there somewhere) I turned all graphical options to the max.


Only when approaching "mach 1" and the aerodynamic effects start to show did it have trouble. I turned those one notch lower than full, and all was well.
Occasionally, when making rapid (that is, really dumb) moves (if your not "testing"), it would stutter.

I dropped the AA down 1 notch as well, as this is the quickest frame rate boost in most games, and I can do pretty much anything I want with a smooth frame rate.


Even flipped it over and landed it, mostly.
Control-able all the way down.



This is on a rocket/ship with 409 parts.


As soon as I find the FPS counter/graph, I'll update with ACTUAL fps #s.



Edit: Most odd combo here. New video card, yet monitor is having trouble.

Priorities!

Edit #2: I put my signature on this one twice! Must be getting old.


Barracuda

Last edited by BarracudaUAK; 05-26-18 at 05:18 AM.
BarracudaUAK is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -5. The time now is 04:32 PM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright © 1995- 2024 Subsim®
"Subsim" is a registered trademark, all rights reserved.