Jump to content

A17 RWG Benchmarking


dsZombiex

Recommended Posts

First go at RWG benchmarking.

EDIT: Observations -

Standardizing seed to 12345!

Seed seems to have a non-trivial impact on time to generate.

GPU loads up during generation, does GPU affect time to generate?

 

----------16384x16384----------

 

dszombiex

9900K

32GB DDR4 3200 16-18-18-36 CR2

b233 hardware change: OCZ RD400 -> Samsung 970 Evo

 

b197

Map 16384x16384

Create: ~ 31 minutes 7 seconds

Load: ~ 5 minutes 49 seconds

b199 seed:12345 <----------Seed seems to have a significant effect on gen time? Or is b199 faster?

Map 16384x16384

Create: ~28 minutes 1 second

Load: ~4 minutes 31 seconds

b233 seed:12345

Map 16384x16384

Create: ~29 minutes 47 seconds

Load: ~50 seconds

 

=Sarsante=

7700K

32GB DDR4 2666

SAMSUNG EVO 850

 

b197

Map 16384x16384

Create: ~45 minutes

Load: ~5 minutes 55 seconds

 

tigerseye

i7-5930k @ 4.62ghz

16GB g.skill 3000mhz 16-17-17-39

Samsung 850 EVO RAID 0

 

b199

Map 16384x16384

Create: ~34 minutes 24 seconds

Load: Not benchmarked

Map 16384x16384 seed:12345

Create: ~29 minutes 48 seconds

Load: CRASH (corrupt generation?)

 

Unknown user

i7-2700k

16GB DDR3 2133 10-12-12-31

Sandisk ultra raid 0

 

b199 seed: 12345

Map 16384x16384

Create: ~44 minutes 27 seconds

Load: CRASH (corrupt generation?)

 

thomas.hock77

AMD FX8120 3.10 GHz

16GB DDR3

OCZ-Vertex3 SSD OS ONLY

ST2000DL003-9Vt166 HDD DATA

 

b199

Map 16384x16384

Create: ~74 mins

Load: ~13 mins

 

Aldakoopa

Ryzen 1700x

16GB DDR4 2800 15-15-15-35

WD Blue 500GB SSD OS ONLY

ST3000VN000 HDD DATA

 

b199 seed: default

Map 16384x16384

Create: ~42 minutes

Load: ~5 minutes

 

NegativeX

Ryzen 2700x (average of 4.1ghz the entire time)

32GB DDR4 3200mhz CL16 Ram

1tb Samsung Evo 860 SSD

 

b199

Map 16394x16384

Create: ~34 minutes 12 second

Load: ~6 minutes 3 seconds

b208 -- RAM OC'd to 3600

Map 16384x16384

Create: ~34 minutes 57 seconds

Load: ~6 minutes 33 seconds

 

levin70

2x E5 2667 v2 Xeons (total of 16 cores 32 threads)

48 gb ECC DDR3

Samsung EVO 860 500gb ssd

GTX 1080

 

b199

Map 16384x16384

Create: ~40 minutes 30 seconds

Load: Not benchmarked

 

Red Eagle LXIX

Intel Xeon W3680 3.325 GHz (25x133) @ 4.20 GHz

24GB G.Skill DDR3 Triple Channel @ 1680 MHz 8-8-8-20 1T

Gigabyte GeForce GTX 1070 8GB

 

b240 seed: RWG size Tests

Map 16384x16384

create ~47 minutes 10 seconds

load ~1 minute 45 seconds

 

----------8192x8192----------

 

NegativeX

Ryzen 2700x

32GB DDR4 3200 16-18-18-36 CR2

Samsung Evo 860

 

b199

Map 8192x8192

Create: ~5 minutes 8 seconds

Load: Not benchmarked

 

dszombiex

9900K

32GB DDR4 3200 16-18-18-36 CR2

b233 hardware change: OCZ RD400 -> Samsung 970 Evo

 

b199

Map 8192x8192

Create: ~ 4 minutes 27 seconds

Load: ~28 seconds

b233 seed:12345

Map 8192x8192

Create: ~ 6 minutes 47 seconds

Load: ~19 seconds

 

Sanktuarium

i7 4790k

32GB DDR3 1600 11-11-11-28 CR2

Adata SU800

 

b199

Map 8192x8192

Create: 4 minutes 37 seconds

Load: 32 seconds

 

tigerseye

i7-5930k @ 4.62ghz

16GB g.skill 3000mhz 17-17-17-39

 

b199

Map 8192x8129 Seed: 12345

Create: ~3 minutes 28 seconds

Load: ~34 seconds

Map 8192x8192 Seed: toospookyformebros

Create: ~2 minutes 57 seconds

Load: ~35 seconds

Link to comment
Share on other sites

I ran the xugenerate on B199 at 16384x16384 and it took 40 minutes 30 seconds approximately

 

It would appear that clock speed more than core count (above 8/16) is important

 

my rig is

 

2x E5 2667 v2 Xeons (total of 16 cores 32 threads)

48 gb ECC DDR3

Samsung EVO 860 500gb ssd

GTX 1080

 

For the first 32 minutes or so (through country roads) i was hitting approx 34% cpu utilization across all threads. Top 16 was running at about 60% and the bottom 16 were running around 20% or less. I was between 9 and 15 gb total usage with the spike occurring at the end for memory usage. Total for my system at the 15 for 7DTD was 19. So depending on what you are doing, that may cause issues if you only have 16 gb of memory. Only one disk usage spike, during export. Never hit 100%. Through road generation, my GPU was averaging close to 40-41% utilization, then after that next to nothing.

Link to comment
Share on other sites

I ran the xugenerate on B199 at 16384x16384 and it took 40 minutes 30 seconds approximately

 

It would appear that clock speed more than core count (above 8/16) is important

 

my rig is

 

2x E5 2667 v2 Xeons (total of 16 cores 32 threads)

48 gb ECC DDR3

Samsung EVO 860 500gb ssd

GTX 1080

 

For the first 32 minutes or so (through country roads) i was hitting approx 34% cpu utilization across all threads. Top 16 was running at about 60% and the bottom 16 were running around 20% or less. I was between 9 and 15 gb total usage with the spike occurring at the end for memory usage. Total for my system at the 15 for 7DTD was 19. So depending on what you are doing, that may cause issues if you only have 16 gb of memory. Only one disk usage spike, during export. Never hit 100%. Through road generation, my GPU was averaging close to 40-41% utilization, then after that next to nothing.

 

Thanks for your help and observations. I've added your data. I was getting ~80% GPU utilization during road generation on a GTX 1080ti. I wonder what it's doing with that GPU.

Link to comment
Share on other sites

Added b208 benchmark by Steam user. User additionally added observations which I will post here:

 

Originally posted by NegativeX:

Decided to test out A17E B208 to see if much has changed but to also monitor RAM/CPU/GPU usage during generation.

 

System Specs:

Ryzen 2700X

32gb DDR4 3200mhz CL16 (OC'd to 3600mhz to see if it helped, which it didn't much)

Geforce GTX 970

512gb Samsung EVO 860

 

Generating a 16k size map took:

34 minutes 57 seconds to generate

6 minutes 33 seconds to load

 

Max RAM usage during generation: 22.4gb's

Max RAM usage after first load up: 11.7gb's (didn't run around, just sat in the starting spot)

Max GPU usage during generation: 17%

Max CPU usage during generation: 35%, but averaged 17%

 

One odd quirk, I had to completely exit out of the game after generating the map, since it never cleared RAM usage and kept adding onto the initial 22.4gb's of use during load up. But restarting the game fixed this.

 

It's my opinion, if you only have 16gb's of ram, forgot about running a 16k map, even if you just download a pregenerated map seperately. (which I think would be possible)

 

Just running on a 16k map, uses almost 12gb's of ram first loading in, which doesn't leave a whole hell of a lot of RAM left over if you only have 16gb's of total system ram. You're likely to start digging into virtual memory at some point during gameplay, which would probably cripple performance.

 

Until they optimize larger maps, if it's even possible with all the new changes, 8k maps are probably the way to go.

Link to comment
Share on other sites

  • 3 weeks later...

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...