Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 2 [3] 4

Author Topic: DF CPU Benchmarking  (Read 10985 times)

BoogieMan

  • Bay Watcher
  • Hi
    • View Profile
Re: DF CPU Benchmarking
« Reply #30 on: November 11, 2019, 07:36:06 am »
RAM is a large contributor to the speed. My extra CPU power over most of these results didn't provide me with much (if any) benefit against the results with better RAM.

Windows 10 Pro x64
All SSDs
Intel Core i7-4790K Devil’s Canyon Quad-Core 4.4GHz (Liquid Cooled)
EVGA GeForce RTX 2070 8GB GDDR6
ASUS SABERTOOTH Z97 MARK 1 USB 3.1
G.SKILL TridentX Series 16GB DDR3 SDRAM 1866 (PC3 14900)

Stable World: 43 seconds
250 years World Gen Complete: 10:04
« Last Edit: November 11, 2019, 07:40:47 am by BoogieMan »
Logged
(╯°□°）╯︵ ┻━┻ BoogieMan, Forumscrub cancels tantrum: Seeking Dr. Pepper

wierd

  • Bay Watcher
  • I like to eat small children.
    • View Profile
Re: DF CPU Benchmarking
« Reply #31 on: November 11, 2019, 07:43:52 am »
You know...  i have this really old fujitsu lifebook e2010 with 1gb of RAM that might be funny to torture...
Logged

methylatedspirit

  • Bay Watcher
  • it/its
    • View Profile
Re: DF CPU Benchmarking
« Reply #32 on: December 16, 2019, 07:41:25 am »
I ran the test on my phone using basically the same Windows emulator that wierd has.
Stable world time: 3 min, 35 s
History Gen to 250: 1 hr, 31 min, 57 s
(Total: 1 hr, 35 min, 32 s)

Specs:
CPU: Exynos 9610 (4 cores @ 2.3 GHz, 4 cores @ 1.6 GHz)
GPU: Mali-G72 MP3
RAM: 6 GB LPDDR4X

Ran the test again for reasons.

Stable world time: 6 min, 21 s
History Gen to 250: 1 hr, 29 min, 21 s
(Total: 1 hr, 35 min, 42 s)
« Last Edit: January 05, 2020, 03:33:43 am by methylatedspirit »
Logged

Überzwerg

  • Bay Watcher
    • View Profile
Re: DF CPU Benchmarking
« Reply #33 on: December 30, 2019, 09:30:38 pm »
~22 minutes total
CPU: Intel Core2Duo E8400 @ 3.00GHz (Launch date Q1'08)
RAM: 4 GB DDR2-800

Not to bad for such an old horse...
Logged

DontMineYellowSnow

  • Bay Watcher
    • View Profile
Re: DF CPU Benchmarking
« Reply #34 on: December 30, 2019, 09:57:08 pm »
Quote from: DontMineYellowSnow on November 02, 2019, 09:31:04 pm
Just ran this:

11 rejected worlds
55 seconds to stable world creation
~8:40 to completion at year 250 (including stable world creation time, so 7:45 spent on history)

I had a couple of browser tabs open and ran the version bundled with the LNP so DFHack, etc. running in the background

Using all the seeds spelled out up top.

Platform:
i7-8700
SSD
16GB DDR4 16-18-18-38 @ ~2880Mhz
ASRock Z390 Phantom Gaming 4 mainboard



anewaname's results seem to suggest that, if the tests were accurate, that extra bump in RAM clock made up for a considerable difference in processing power.  What kind of RAM you got there?  Specs on latency?  Are you using standard XMP to get to 3000Mhz?

Just to follow up, I reran the exact same thing after bumping my RAM to 3000 with the same timings:

~52 seconds to stable world creation
~8:20 to worldgen completion (including stable world creation time, so 7:28 spent on history

Thats a tiny bump for noticeable performance differences.  Any other program wouldn't even notice such a bump.


Next step, I plan to get some of that higher end 4000Mhz RAM for myself and swap it in.  Then, when the 10 series chips release with their 20MB L3 SmartCache, I'm going to upgrade the CPU in this computer.  I'll retry incrementally with the new hardware and will post back as they become available.
Logged

methylatedspirit

  • Bay Watcher
  • it/its
    • View Profile
Re: DF CPU Benchmarking
« Reply #35 on: January 04, 2020, 04:37:29 pm »
Tested another phone, a Google Pixel 2.
Stable world time: 3 min, 26 s
History Gen to 250: 1 hr, 53 min, 39 s
(Total: 1 hr, 57 min, 5 s)

Specs:
CPU: Qualcomm Snapdragon 835 (4 cores @ 2.45 GHz, 4 cores @ 1.9 GHz)
GPU: Adreno 540
RAM: 4 GB LPDDR4X

It's a bit of a surprise that it takes longer than the last phone (a Galaxy A50), given that it slightly outperforms the A50 in CPU benchmarks.

EDIT: Tried it again to check if it was a fluke. Got pretty much the same results:
Stable world time: 3 min, 20 s
History Gen to 250: 1 hr, 52 min, 27 s
(Total: 1 hr, 55 min, 47 s)
« Last Edit: January 04, 2020, 06:52:05 pm by methylatedspirit »
Logged

wierd

  • Bay Watcher
  • I like to eat small children.
    • View Profile
Re: DF CPU Benchmarking
« Reply #36 on: January 04, 2020, 11:37:46 pm »
Might be memory speed limited rather than CPU speed limited.  Remember, emulation does much more work than native operation, and that includes more RAM accesses.
Logged

methylatedspirit

  • Bay Watcher
  • it/its
    • View Profile
Re: DF CPU Benchmarking
« Reply #37 on: January 05, 2020, 12:19:59 am »
Quote from: wierd on January 04, 2020, 11:37:46 pm
Might be memory speed limited rather than CPU speed limited.  Remember, emulation does much more work than native operation, and that includes more RAM accesses.
RAM doesn't seem to be it. The clock of the memory controller of the Pixel 2 is 1866 MHz as compared to 1600 MHz of the A50. I also ran a memory benchmark (CPDT), and the Pixel 2 got 6.15 GB/s in memory copy compared to 3.56 GB/s of the A50. The Pixel seems to have faster RAM than the A50, yet it gets beaten by the A50 when emulating DF.
« Last Edit: January 05, 2020, 12:29:24 am by methylatedspirit »
Logged

wierd

  • Bay Watcher
  • I like to eat small children.
    • View Profile
Re: DF CPU Benchmarking
« Reply #38 on: January 05, 2020, 12:55:06 am »
Odd...  No idea.
Logged

methylatedspirit

  • Bay Watcher
  • it/its
    • View Profile
Re: DF CPU Benchmarking
« Reply #39 on: January 05, 2020, 03:54:43 am »
I thought of something: thermal throttling. My theory is that the Pixel, while slightly more powerful than the A50 when cold, produces more heat than the A50 (or is more aggressive with thermal throttling), forcing it to throttle the CPU more, thus yielding worse sustained performance than the A50.

So I ran a thermal throttling benchmark on both phones for 30 minutes. The A50 throttled to 87% of its max performance, while the Pixel throttled to 74% of its maximum performance.

I'm not sure how that explains how the Pixel took a whole half hour longer than the A50.
« Last Edit: January 05, 2020, 07:16:59 am by methylatedspirit »
Logged

anewaname

  • Bay Watcher
  • The mattock... My choice for problem solving.
    • View Profile
Re: DF CPU Benchmarking
« Reply #40 on: January 05, 2020, 03:45:05 pm »
One problem with testing DF on this device or that device, is that there is no way to determine what the relative value of other potential bottlenecks are.

If you increased memory speed, did the low benefit to DF's speed indicate that the bottleneck is now at the motherboard or the cpu, blocking some of the benefits of the faster memory?

How are speed ratings for cpu and memory effected by a phone's motherboard design features (features that were added to benefit certain types of processing such as video-rendering and display)?
Logged
Quote from: dragdeler
There is something to be said about, if the stakes are as high, maybe reconsider your certitudes. One has to be aggressively allistic to feel entitled to be able to trust. But it won't happen to me, my bit doesn't count etc etc... Just saying, after my recent experiences I couldn't trust the public if I wanted to. People got their risk assessment neurons rotten and replaced with game theory. Folks walk around like fat turkeys taunting the world to slaughter them.

methylatedspirit

  • Bay Watcher
  • it/its
    • View Profile
Re: DF CPU Benchmarking
« Reply #41 on: January 06, 2020, 12:34:26 am »
Quote from: anewaname on January 05, 2020, 03:45:05 pm
One problem with testing DF on this device or that device, is that there is no way to determine what the relative value of other potential bottlenecks are.

If you increased memory speed, did the low benefit to DF's speed indicate that the bottleneck is now at the motherboard or the cpu, blocking some of the benefits of the faster memory?

How are speed ratings for cpu and memory effected by a phone's motherboard design features (features that were added to benefit certain types of processing such as video-rendering and display)?
Good point, but it's not gonna stop me from benchmarking every phone I can get my hands on.

Anyway, I tested a Huawei Mate 10 Pro.
Stable world time: 13 min, 22 s
History Gen to 250: 4 hrs, 29 min, 38 s
(Total: 4 hr, 43 min)

Specs:
CPU: HiSilicon Kirin 970 (4 cores @ 2.36 GHz, 4 cores @ 1.8 GHz)
GPU: Mali-G72
RAM: 6 GB LPDDR4X

I won't even pretend to know what happened here.

EDIT: Tried it again, this time with the phone in Performance mode.
Stable world time: 3 min, 4 s
History Gen to 250: 1 hr, 13 min, 24 s
(Total: 1 hr, 16 min, 28 s)
« Last Edit: January 06, 2020, 01:58:17 am by methylatedspirit »
Logged

Knas

  • Escaped Lunatic
    • View Profile
Re: DF CPU Benchmarking
« Reply #42 on: January 07, 2020, 08:37:43 am »
Linux Manjaro
CPU: AMD Ryzen 7 3700X
RAM: 16GB (2x8GB) DDR4 3733MHz CL16

Stable world: 16.85s
Year 250: 7min 21.97s
(Total: 7min 38.83s)

I'm getting significantly slower times on Win 10 VM running on the same machine (10min+ total). I should try booting up native Windows 10 and see if it's any faster.

Edit:
Windows 10 (not vm)
(same exact hardware)

Stable world: 27s
Year 250: 9min 4s
(Total: 9min 31s)
« Last Edit: January 08, 2020, 07:19:28 am by Knas »
Logged

methylatedspirit

  • Bay Watcher
  • it/its
    • View Profile
Re: DF CPU Benchmarking
« Reply #43 on: January 29, 2020, 07:04:18 am »
47.01 just came out, which means it's time to redo the benchmark (using the same seeds) on the new version.

Stable world time: 56 seconds
History Gen to 250: 18 minutes, 12 seconds
(Total: 19 minutes, 8 seconds)

DF Version: 47.01
CPU: Intel Core i5-4460 @ 3.20 GHz
GPU: AMD Radeon R7 200
RAM: 8 GB @ 1600 MHz

Spoiler: Worldgen config used (click to show/hide)
Stable world time: 2 min, 41s
History Gen to 250: 34 min, 17s
(Total: 36 min, 58s)

DF Version: 47.01
CPU: AMD A9-9420, 2 cores @ 3.40 GHz
GPU: Radeon 520
RAM: 8 GB @ 1866 MHz

Stable world time: 3 min, 40 s
History Gen to 250: 3 hr, 14 min, 11 s
(Total: 3 hr, 17 min, 51 s)

DF Version: 47.01
CPU: Qualcomm Snapdragon 835 (4 cores @ 2.45 GHz, 4 cores @ 1.9 GHz)
GPU: Adreno 540
RAM: 4 GB LPDDR4X
Stable world time: 5 min, 53 s
History Gen to 250: 1 hr, 51 min, 23 s
(Total: 1 hr, 57 min, 16 s)

DF Version: 47.01
CPU: Exynos 9610 (4 cores @ 2.3 GHz, 4 cores @ 1.6 GHz)
GPU: Mali-G72 MP3
RAM: 6 GB LPDDR4X
« Last Edit: January 30, 2020, 01:27:56 am by methylatedspirit »
Logged

methylatedspirit

  • Bay Watcher
  • it/its
    • View Profile
Re: DF CPU Benchmarking
« Reply #44 on: July 22, 2021, 09:22:05 am »
Went back to 44.12 for this one. This is a Raspberry Pi 4B, running Linux DF under Box86 with Dynarec v0.2.3, Raspbian Buster armhf.

Stable world time: 8 minutes, 5 seconds
History Gen to 250: 1 hour, 0 minutes, 26 seconds
(Total: 1 hour, 8 minutes, 31 seconds)

CPU: Broadcom BCM2711, Quad core Cortex-A72 @ 2.0 GHz (overclocked from 1.5)
GPU: Broadcom VideoCore VI @ 500 MHz
RAM: 4 GB LPDDR4-3200
Logged
Pages: 1 2 [3] 4
« previous next »