Benchmark Testing of Spaced out

[FGPraxis]

@Francis John : (after watching your video 27-09 and reading all these posts/results)

IMO you should go for 2x16Go DDR4-4000CL16-16-16-36@1.4V (or 4266)

But don't buy the first DDR4-4000 you see, for example Gskill have three series :

Trident Z Neo : your pick, it's optimized for AMD RYZEN (F4-4000C16D-32GTZNA)

Trident Z royal : another good pick, it's compatible/optimized for both INTEL and AMD (F4-4000C16D-32GTRGA or F4-4000C16D-32GTRSA)

RipjawsV : avoid it, it's optimized for INTEL

Most of the time DDR4 manufacturers test and guarantee compatibility with a list of Motherboards, check it, it's important if you want a reliable/stable high perf CPU+memory core.

[gabberworld]

17 minutes ago, FGPraxis said:

@Francis John : (after watching your video 27-09 and reading all these posts/results)

IMO you should go for 2x16Go DDR4-4000CL16-16-16-36@1.4V (or 4266)

But don't buy the first DDR4-4000 you see, for example Gskill have three series :

Trident Z Neo : your pick, it's optimized for AMD RYZEN (F4-4000C16D-32GTZNA)

Trident Z royal : another good pick, it's compatible/optimized for both INTEL and AMD (F4-4000C16D-32GTRGA or F4-4000C16D-32GTRSA)

RipjawsV : avoid it, it's optimized for INTEL

Most of the time DDR4 manufacturers test and guarantee compatibility with a list of Motherboards, check it, it's important if you want a reliable/stable high perf CPU+memory core.

why cl16 ? why not cl15?

[FGPraxis]

2 minutes ago, gabberworld said:

why cl16 ? why not cl15?

Why not yes,

I just looked at GSkill, i don't know about other manufacturers, these lists are very long to check

I didn't exactly see 4000CL15, it's either 4000CL16-16-16-36 (very good), or 4000CL16-19-19-39 (good), or 4000CL14-15-15-35 (very good top notch, can probably be considered CL15 but referenced as CL14)

Then i checked the prices on my favorite eshop and i only saw the 4000CL16-16-16-36 and the 4000CL16-19-19-39, not the CL14

If the 4000CL14 is not overpriced, why not yes

[gabberworld]

1 minute ago, FGPraxis said:

Why not yes,

I just looked at GSkill, i don't know about other manufacturers, these lists are very long to check

I didn't exactly see 4000CL15, it's either 4000CL16-16-16-36 (very good), or 4000CL16-19-19-39 (good), or 4000CL14-15-15-35 (very good top notch, can probably be considered CL15 but referenced as CL14)

Then i checked the prices on my favorite eshop and i only saw the 4000CL16-16-16-36 and the 4000CL16-19-19-39, not the CL14

If the 4000CL14 is not overpriced, why not yes

they have 4000 cl15 . i not seen 4000 cl14 yet

[FGPraxis]

2 minutes ago, gabberworld said:

they have 4000 cl15 . i not seen 4000 cl14 yet

huh ? i'm on Gskill site : F4-4000C14D-32GTZN or F4-4000C14D-32GTRS or F4-4000C14D-32GTRG, it's not important but it's CL14, i don't see a CL15

I've found a price for these, it's half more, but why not

[Orical019]

A little less relevant to all the other results because I play on a laptop but I wanted to join in anyways. I have a ASUS ROG G17 with upgraded memory. Benchmark results were run with fans on turbo and no other programs active, but I also ran it in the method I usually play the game (searching matchmaking in another game with the fans ramped down to a more acceptable level) and it's time was nearly identical (~1s difference) Looking forward to a bit of breakdown on all this data!

Time : 146s (2:26)
CPU : Ryzen 9 5900HX
Graphics card : RTX 3070 (laptop)
RAM : 32GB 3200mhz

CAS : 22,22,22,52 (wish ASUS allowed me to load the XMP profile but this model laptop doesn't allow any overclocking)
HDD : ADATA XPG S7 2TB

[gabberworld]

3 minutes ago, FGPraxis said:

huh ? i'm on Gskill site : F4-4000C14D-32GTZN or F4-4000C14D-32GTRS or F4-4000C14D-32GTRG, it's not important but it's CL14, i don't see a CL15

I've found a price for these, it's half more, but why not

15-16-16-36. but i found also cl14 now. man yeah that price is high

[amabitxs]

I don't know that my results will help much as I too need a computer upgrade... but here they are!

Time: 3:14 (194s)

CPU: Intel Core i7 8700K

GPU: NVIDIA GeForce GTX 1050

RAM: 16GB 2133MHz

CAS: 15/15/15/36

HHD: WDC WDS512G1X0C-00ENX0

Cooler: Water cooled CPU, Fan cooled GPU.

 

Keep up the good work!

[Squinch]

Time : 2:05
CPU : Intel i9-10850k CPU @ 3.60GHz
Graphics card : NVIDIA GeForce RTX 3080
RAM :32 GB

[Durandal337]

Time: 2:22:59 142.59 sec
CPU: Core i7-9700F CPU@3.00GHz (approx 4.4GHz)
Graphics Card: NVIDIA GEForce RTX 2060 Super 8GB
RAM: G.SKILL RipJaws V DDR4 2133Mhz 32GB (8GB x4) 
HDD: Samsung 1TB 870 EVO 2.5in SATA SSD
Display: Gigabyte G34WQC  3440x1440 @144hz 
MB: ROG Strix B360-F Gaming

Not the newest parts I'll admit just the last of the previous generation. CPU stayed at about 35-37% and Graphics at a bit over 9GB. I had the same stutter at about 70ish seconds but otherwise everything was fairly smooth at fullscreen.

[Bearded Dwarf]

Time : 3:23 (203 seconds)
CPU : 2.3 GHz Intel Core i5
Graphics card : Intel Iris Plus Graphics 640 1536 MB
RAM : 16 GB 2133 MHz LPDDR3

CAS : 
HDD : APPLE SSD AP0256J

This is a 2017 13in Macbook Pro so feel free to exclude this from your results if you want. It does stack up very well for the i5 though  

[gabberworld]

11 minutes ago, amabitxs said:

 computer upgrade

 

mostly what i see in this test is that people need memory upgrade

[FGPraxis]

12 hours ago, gabberworld said:

15-16-16-36. but i found also cl14 now. man yeah that price is high

Yes at this point the added performance/added cost isn't good, better go for 4266, because memory frequency >= memory latency

The only question that remains is the compatibility, i'm surprised Gigabyte is not listed for the 4000CL16 i advise, i bet they are running the tests and will update the list soon (Gigabyte is very good, it's my 2nd pick after MSI for MB)

Btw it may look like flooding the thread, but after reading all the results, it's clear a lot of config are RAM limited, either quantity (16 is min 32 is safe) or frequency (3600 is min, but what is the max safe 4000, 4266, 5000 lol ?) or latency (these 3600CL19 don't look good)

Because CPU frequency is very limited indeed even for pro oc (there is a good reason CPU manufacturers are going multicore a lot), Memory frequency and low latency is where we can push performance, people should focus more on it

Other than that i'll let the pro overclockers give more advises about max frequencies and compabilities

Good luck and fun !

[tyraelpl]

23 minutes ago, amabitxs said:

I don't know that my results will help much as I too need a computer upgrade... but here they are!

Time: 3:14 (194s)

CPU: Intel Core i7 8700K

GPU: NVIDIA GeForce GTX 1050

RAM: 16GB 2133MHz

CAS: 15/15/15/36

HHD: WDC WDS512G1X0C-00ENX0

Cooler: Water cooled CPU, Fan cooled GPU.

 

Keep up the good work!

Oh this looks like a serious case of thermal throttling if ive ever seen one! Worse a bit but near same range result than my 3570k @ 4,2 GHz with ddr3 ram (@1600). I had like ~190 s. Wow!

[Tuberi]

1 hour ago, amabitxs said:

I don't know that my results will help much as I too need a computer upgrade... but here they are!

Time: 3:14 (194s)

CPU: Intel Core i7 8700K

GPU: NVIDIA GeForce GTX 1050

RAM: 16GB 2133MHz

CAS: 15/15/15/36

HHD: WDC WDS512G1X0C-00ENX0

Cooler: Water cooled CPU, Fan cooled GPU.

 

Keep up the good work!

Your memory is probably rated for higher speed than what they are set to right now so enabling XMP in bios might give you a pretty big boost

 

57 minutes ago, tyraelpl said:

Oh this looks like a serious case of thermal throttling if ive ever seen one! Worse a bit but near same range result than my 3570k @ 4,2 GHz with ddr3 ram (@1600). I had like ~190 s. Wow!

I don't think ONI is capable of causing thermal throttling outside of a dual core CPU or a CPU with no cooler attached

[gabberworld]

1 hour ago, Durandal337 said:

Time: 2:22:59 142.59 sec
CPU: Core i7-9700F CPU@3.00GHz (approx 4.4GHz)
Graphics Card: NVIDIA GEForce RTX 2060 Super 8GB
RAM: G.SKILL RipJaws V DDR4 2133Mhz 32GB (8GB x4) 
HDD: Samsung 1TB 870 EVO 2.5in SATA SSD
Display: Gigabyte G34WQC  3440x1440 @144hz 
MB: ROG Strix B360-F Gaming

Not the newest parts I'll admit just the last of the previous generation. CPU stayed at about 35-37% and Graphics at a bit over 9GB. I had the same stutter at about 70ish seconds but otherwise everything was fairly smooth at fullscreen.

you could probably also increase speed by change the motherboard and memory

example the MSI. Z390-A PRO

[FryedWater]

Time : 2:35 (255)
CPU : ryzen 5 3600 with PBO on
Graphics card : GTX 1660 super
RAM : 16gb 3200MHz 2x8gb

CAS : 16-18-18-36
HDD : windows is on Kingston A400 240 GB, game is on WD Blue 1 TB 7200RPM

https://au.pcpartpicker.com/list/DKLQXb full list of specs

[DeadlyPunker]

4 hours ago, DeadlyPunker said:

Time: 1:55 (115 seconds)

CPU: Ryzen 5800x

Graphics Card: GTX 1070 8gb

Ram:  32GB (2 x 16GB) DDR4 4000 (PC4 32000)

CAS: 16-19-19-39

 

Following up

I got the following with changes to my ram

Time 1:51 (111 seconds)

Ram:  32GB (2 x 16GB) 3600

CAS 16-16-16-36

 

Then I furthered tightened my settings to:

Time 1:51 (111 seconds)

Ram:  32GB (2 x 16GB) 4000

CAS 16-16-16-34

I am ran the test a couple times and was seeing about +/- 1 second, restart ONI each run.  There appears to a pretty steep curve of diminishing returns for the ranges I am testing.

[gabberworld]

2 hours ago, DeadlyPunker said:

Following up

I got the following with changes to my ram

Time 1:51 (111 seconds)

Ram:  32GB (2 x 16GB) 3600

CAS 16-16-16-36

 

Then I furthered tightened my settings to:

Time 1:51 (111 seconds)

Ram:  32GB (2 x 16GB) 4000

CAS 16-16-16-34

I am ran the test a couple times and was seeing about +/- 1 second, restart ONI each run.  There appears to a pretty steep curve of diminishing returns for the ranges I am testing.

accounting to gskill 16-16-16-36 is proper 4000 for some off they selled memorys. 16-16-16-34 may not give any extra effect as they already tested those out

 

[DeadlyPunker]

17 minutes ago, gabberworld said:

accounting to gskill 16-16-16-36 is proper 4000 for some off they selled memorys. 16-16-16-34 may not give any extra effect as they already tested those out

 

I changed the clock too.  A 10% under clock is not small.  But the point that i observed was no gain from the frequency change.  However timing changes did get me 4 seconds.

[gabberworld]

11 minutes ago, DeadlyPunker said:

I changed the clock too.  A 10% under clock is not small.  But the point that i observed was no gain from the frequency change.  However timing changes did get me 4 seconds.

that is 16-19-19-39 some off they different selled ones. i quess its good that you try different settings.

and yes 4 sec gain is very good

[Kerdeld]

11 hours ago, gabberworld said:

ok. what company memory you use?

also can you share your HWiNFO image

@gabberworld 

https://imgur.com/a/1FbNzbB Hwinfo, I just woke up so pc have been on for only 11mins if want anything else lemme know 

 

9 hours ago, Krunkkracker said:

Hey there everyone! I've actually done this before and made a Reddit post about it. I built this system with ONI and other simulations like X4 Foundations, Dwarf Fortress, Aorura4x, and others in mind. If anyone wants further details about my subtimings, voltages, etc just ask! Also I can do a capture came recording/livestream in order to prove the 103. I still have some headroom left on my ram, I could probably get sub 100s with daily safe and stable settings if I took a day or two to tune my ram subtimings. 

Time : 103
CPU : 10900kf@5.2ghz/4.9ghz core/ring and HT off
Graphics card : MSI Ventus LHR RTX 3060 ti
RAM : 2x16gb 4500mhz Dual Rank

CAS : 16-17-17-34
HDD : Adata SX8200 Pro 1tb

https://old.reddit.com/r/Oxygennotincluded/comments/hcdwpv/performance_testing_and_scaling_with_cycle_time/

@Krunkkracker can I have a Zentimings picture and DIMM voltage that must be crazy high or the kit was really expensive

8 hours ago, FGPraxis said:

PC custom 8 months old

Time : 2:00 - 120sec
CPU : i7-10700F running@4.6Ghz HyperThreading on (55°C during test)
GC : MSI GTX1060
RAM : 2x8Go DDR4-3200 RipjawsV (XMP on)
CAS : 14-14-14-34
HDD : SSD 860EVO 250Go (Win10Pro64) + SSD 860EVO 1To (games) + hdd Samsung 2To
MB MSI Z490 GP

Time : 1:57 - 117sec
CPU : i7-10700F running@4.6Ghz HyperThreading off (55°C during test)
GC : MSI GTX1060
RAM : 2x8Go DDR4-3200 RipjawsV (XMP on)
CAS : 14-14-14-34
HDD : SSD 860EVO 250Go (Win10Pro64) + SSD 860EVO 1To (games) + hdd Samsung 2To
MB MSI Z490 GP

Sidenotes :
- i didn't have a lot of occasions to test with/without HT, thanks to this test i know i benefit a little from HT off, thanks FJ

- this save is very laggy compared to my current 4500 cycles game that have almost no lag or stutter, but i don't have 3000 pacus, and the two large planetoids (main and 2nd) are almost entirely filled with oxygen and/or clean water

- Max frequency of CPU 1core/4cores and RAM (and low CAS) is the key i guess

- my CPU is cooled by a Thermalright TrueSpirit 120 Direct A + BeQuiet PureWing2 : i consider silent even at full charge

@FGPraxis If you learn how to overclock the ram, it will likely hit really high bandwidths and offer a lot more performance for you

5 hours ago, amabitxs said:

I don't know that my results will help much as I too need a computer upgrade... but here they are!

Time: 3:14 (194s)

CPU: Intel Core i7 8700K

GPU: NVIDIA GeForce GTX 1050

RAM: 16GB 2133MHz

CAS: 15/15/15/36

HHD: WDC WDS512G1X0C-00ENX0

Cooler: Water cooled CPU, Fan cooled GPU.

 

Keep up the good work!

 @amabitxs is your ram specs 2133 or XMP not turned on?

5 hours ago, Durandal337 said:

Time: 2:22:59 142.59 sec
CPU: Core i7-9700F CPU@3.00GHz (approx 4.4GHz)
Graphics Card: NVIDIA GEForce RTX 2060 Super 8GB
RAM: G.SKILL RipJaws V DDR4 2133Mhz 32GB (8GB x4) 
HDD: Samsung 1TB 870 EVO 2.5in SATA SSD
Display: Gigabyte G34WQC  3440x1440 @144hz 
MB: ROG Strix B360-F Gaming

Not the newest parts I'll admit just the last of the previous generation. CPU stayed at about 35-37% and Graphics at a bit over 9GB. I had the same stutter at about 70ish seconds but otherwise everything was fairly smooth at fullscreen.

 @Durandal337 can you check your ram in CPU-z

like this and see for the most right timing table, does it say the same like mine? 1600 14 14 14 34 48?

4 hours ago, gabberworld said:

mostly what i see in this test is that people need memory upgrade

I think more important is overclocking the Ram for higher bandwidth and often 3200+ with tight manual timings help a lot and overclocking the CPU by like 100-300 MHz depends on the CPU and how you do it.

Personally I don't think 3800MHz+ ram is worth it, it's often a lot more pricey if timings are fine and not always guaranteed to work out of the box depending on the CPU.

I had a Ryzen 2600 before that would do 2x8 3400cl14 at the most and anything on 3466 even cl16 or 18 would crash 

then I bought another 2x8 and max I could run was 3200cl14 tight timings 

With the 5600x the limit was totally gone and around 3800+ you can find a lot of problems if you don't know a lot about ram and 4000+ again even more if your ram can handle it. 

For Intel this is a different story, 4000+ is handled better there like we see with @Krunkkracker

 

2 hours ago, DeadlyPunker said:

Following up

I got the following with changes to my ram

Time 1:51 (111 seconds)

Ram:  32GB (2 x 16GB) 3600

CAS 16-16-16-36

 

Then I furthered tightened my settings to:

Time 1:51 (111 seconds)

Ram:  32GB (2 x 16GB) 4000

CAS 16-16-16-34

I am ran the test a couple times and was seeing about +/- 1 second, restart ONI each run.  There appears to a pretty steep curve of diminishing returns for the ranges I am testing.

Yup the same for me at 3733 the gains stopped, it can again I guess above 4200+ it can pick up again but I think it's most likely only the Intel CPU's who can do it because of the way AMD handles infinity fabric overclocking and combability with motherboard/CPU/Ram makes it a lot harder.

 

[gabberworld]

Quote

I think more important is overclocking the Ram for higher bandwidth and often 3200+ with tight manual timings help a lot and overclocking the CPU by like 100-300 MHz depends on the CPU and how you do it.

Personally I don't think 3800MHz+ ram is worth it, it's often a lot more pricey if timings are fine and not always guaranteed to work out of the box depending on the CPU.

I had a Ryzen 2600 before that would do 2x8 3400cl14 at the most and anything on 3466 even cl16 or 18 would crash 

then I bought another 2x8 and max I could run was 3200cl14 tight timings 

With the 5600x the limit was totally gone and around 3800+ you can find a lot of problems if you don't know a lot about ram and 4000+ again even more if your ram can handle it. 

For Intel this is a different story, 4000+ is handled better there like we see with @Krunkkracker

yes @Kerdeld it crash that why there need find also proper motherboard.  some may just not work

[Tuberi]

4 hours ago, DeadlyPunker said:

Following up

I got the following with changes to my ram

Time 1:51 (111 seconds)

Ram:  32GB (2 x 16GB) 3600

CAS 16-16-16-36

 

Then I furthered tightened my settings to:

Time 1:51 (111 seconds)

Ram:  32GB (2 x 16GB) 4000

CAS 16-16-16-34

I am ran the test a couple times and was seeing about +/- 1 second, restart ONI each run.  There appears to a pretty steep curve of diminishing returns for the ranges I am testing.

Did you confirm your ram speeds were actually what you set them to?

[electrifier]

Time : 2:57
CPU : i7-5820K @ 3.3GHz
Graphics card : GTX 1080Ti
RAM : 32GB DDR4 @ 2800MHz 

CAS : 16/18/18/36
HDD : Crucial CT1000P1SSD8 (1Tb M.2 NVME)