Forged Alliance Forever Forged Alliance Forever Forums 2020-03-07T04:06:09+02:00 /feed.php?f=2&t=18816 2020-03-07T04:06:09+02:00 2020-03-07T04:06:09+02:00 /viewtopic.php?t=18816&p=182422#p182422 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
ZLO_RD wrote:
Edit: lol this test is kinda meaningless xD. 3000 beats 3200 with same timings. 2133 has good timings and no latency data, so can't compare results.


Yes, this happens if ppl dont know how a PC works and try to overclock.
If you change the speed of your memory then you also need to set the speed from your front side bus (mainboard clock)
This is needed to set the clock for the mainboards north bridge.

North Bridge is the memory manager/controller:
Bridge.png

Unsync memory/NorthBridge clock settings will lead to very low memory latency:
You can see here that you will lose about 20% memory speed (Additional Async Latency Offset)

Async.jpg

So its no surprise to have fast benchmark results with 3000 clock because 2133 and 3200 are set unsync to North Bridge.
Have in mind we need fast memory "latency" for SupCom, we don't need a high memory clock or high memory throughput.

Statistics: Posted by Uveso — 07 Mar 2020, 04:06


]]>
2020-03-06T15:34:52+02:00 2020-03-06T15:34:52+02:00 /viewtopic.php?t=18816&p=182420#p182420 <![CDATA[Re: How good Athlon200GE would be in SupCom]]> https://www.youtube.com/watch?v=9qYT6GrnK38&t=22s
someone made a test here
timings are in the comments
Edit: lol this test is kinda meaningless xD. 3000 beats 3200 with same timings. 2133 has good timings and no latency data, so can't compare results.

Statistics: Posted by ZLO_RD — 06 Mar 2020, 15:34


]]>
2020-03-06T16:16:36+02:00 2020-03-06T10:51:57+02:00 /viewtopic.php?t=18816&p=182417#p182417 <![CDATA[Athlon200GE in SupCom anyone? (memory performance importance]]> i did 2 runs.
run 1 22:56
run 2 26:07

Statistics: Posted by ZLO_RD — 06 Mar 2020, 10:51


]]>
2020-03-06T10:49:08+02:00 2020-03-06T10:49:08+02:00 /viewtopic.php?t=18816&p=182416#p182416 <![CDATA[Re: How good Athlon200GE would be in SupCom]]> I overclocked my memory a little bit and lowered timings.
But Xeon E5 1650 right now is set up with energy-saving settings all "on", and instead of using less CPU it just drops clocks. When OCed to 3.9 it drops to 3.7 ~at 10 minute in replay. 3.1 at 25+ minutes... As if there is nothing to load CPU with.
i use time to go trought a replay as a benchmark and from two runs i got 3 minutes difference, pretty undstable.
i will change settings and do more tests later.
We use 11061063 replay to test.

Some other people's tests:

21:37 (ryzen 2600 4.03 Ггц mobo B450+ 16Гб ram 2933 Мгц)

i7-4770k, CPU rate 147, CPU clock jumps around 4,0-4,2, DDR3 2200 MHz
watched for 19:46, still +0 at the end of the game.

1) ryzen 3600, CPU rate 145,frequency fixed at 4,0, DDR4 3600 MHz
watched for 13:16, +2 at the end of the game.
2) 3000 ram 14:36 replay time
3) 2400 ram 16:18 replay time
+ aida screenshots

Statistics: Posted by ZLO_RD — 06 Mar 2020, 10:49


]]>
2020-03-05T23:03:15+02:00 2020-03-05T23:03:15+02:00 /viewtopic.php?t=18816&p=182413#p182413 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
ZLO_RD wrote:
i thought they mean 1 core is running at 30% :?


Yes, all cores are under 30% load.
So a single thread could use 70% more CPU power if needed.

Statistics: Posted by Uveso — 05 Mar 2020, 23:03


]]>
2020-03-04T20:21:48+02:00 2020-03-04T20:21:48+02:00 /viewtopic.php?t=18816&p=182402#p182402 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
TheKoopa wrote:
The cpu is running at 30% because most supcom action happens on a single thread

The gpu does not affect sim speed

i thought they mean 1 core is running at 30% :?

Statistics: Posted by ZLO_RD — 04 Mar 2020, 20:21


]]>
2020-03-04T15:05:35+02:00 2020-03-04T15:05:35+02:00 /viewtopic.php?t=18816&p=182396#p182396 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
The gpu does not affect sim speed

Statistics: Posted by TheKoopa — 04 Mar 2020, 15:05


]]>
2020-03-03T20:10:09+02:00 2020-03-03T20:10:09+02:00 /viewtopic.php?t=18816&p=182384#p182384 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
ZLO_RD wrote:
correct me if i am wrong, but amount of channels and memory rank do not improve latency.


Thats absolutly right. This will only increase the datarate like RAID 0 for HDD.

Statistics: Posted by Uveso — 03 Mar 2020, 20:10


]]>
2020-03-03T15:27:18+02:00 2020-03-03T15:27:18+02:00 /viewtopic.php?t=18816&p=182380#p182380 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
it really sounds like key here is latency and not memory bandwidth.

correct me if i am wrong, but amount of channels and memory rank do not improve latency.

Anyway... need someone esle to test this too. i will try to test myself. can't beleave that you got same result at ~2Ghz. Internal caches and stuff like that often depends on CPU frequency and dropping frequency from 4 to 2 Ghz should also reduce memory-system performance.

Statistics: Posted by ZLO_RD — 03 Mar 2020, 15:27


]]>
2020-03-03T09:48:05+02:00 2020-03-03T09:48:05+02:00 /viewtopic.php?t=18816&p=182378#p182378 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
since we have fast CPUs and GPUs many players have already mentioned that the simspeed is low even when their CPUs/GPUs are only running with 30% load.
So there must be another bottleneck that causes the low simspeed.

So i started some RAM testing to see if maybe the RAM is causing the low simspeed.
I have 5 PCs here for testing; i7-4790K, i7-6700K, Ryzon 5, and 2x intel Core2 Quad @ 2.5 and 2.6 Ghz.

I increased or decreased first the CPU clock with almost no effect on the simspeed.
Even when i decrease the CPU clock from 4Ghz down to 1.8 Ghz, the simspeed still does not go slower.

Then i tested the GPU. There was no effect at all.

Last i tested RAM speed.
Effect: if you decrease the RAM clock then also the simspeed is going slower. And in case you increase RAM speed then also the simspeed is faster.
I did some additional checks with diferent RAS-to-CAS delay, CAS latency and Row Precharge and Cycle Times.

When you buy RAM modules you will see numbers like "CL8 8-8-24"
That means:
CL8 = CAS Latency
8 = RAS to CAS delay.
8 = RAS precharge
24 = active to precharge Delay
(If you buy RAM; lower numbers are better!)

Most important is here 24 = active to precharge delay.
This is the time a RAM needs to internally refresh the row where we have written the last data.

Also important, use RAM that are doublesided. (mostly they have momorychips on both sides of the memory module)
And try to use a mainboard that can use memory in quadchannel, at least dualchannel.

I am telling this since 2 years here, but no one cares ^^
2 year old post: forums/viewtopic.php?f=45&t=15908#p160934

Statistics: Posted by Uveso — 03 Mar 2020, 09:48


]]>
2020-03-02T18:53:24+02:00 2020-03-02T18:53:24+02:00 /viewtopic.php?t=18816&p=182369#p182369 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
Uveso wrote:
WAY more important, what RAM are you plan to use ?
Faster RAM = faster sim speed.
I would recommend DDR4 over DDR3 RAM.
Also better use RAM with a low latency than a high DATA rate.


How is that so important? There is no infinity fabric between cores, it is only between CPU and GPU, so RAM will probably affect FPS even more than the sim speed...
You can only use DDR4 on AM4 obviosly... Since goal is to get cheapest PC i think you should buy cheapest or best value RAM... some cheapest crucial or samsung, some OEM stuff... i don't really know much about RAM. Also i thought only powerfull CPUs are really limited by RAM speed... but i guess in our case you still want speed and good timings for lower altency... imho RAM is very complicated and might not worth buying something premium consideting we use Athlon 200GE and not even an ryzen 1600...

Statistics: Posted by ZLO_RD — 02 Mar 2020, 18:53


]]>
2020-03-02T18:36:01+02:00 2020-03-02T18:36:01+02:00 /viewtopic.php?t=18816&p=182368#p182368 <![CDATA[Re: How good Athlon200GE would be in SupCom]]>
WAY more important, what RAM are you plan to use ?
Faster RAM = faster sim speed.
I would recommend DDR4 over DDR3 RAM.
Also better use RAM with a low latency than a high DATA rate.

Statistics: Posted by Uveso — 02 Mar 2020, 18:36


]]>
2020-03-02T15:58:34+02:00 2020-03-02T15:58:34+02:00 /viewtopic.php?t=18816&p=182365#p182365 <![CDATA[How good Athlon200GE would be in SupCom]]> I used to play few games on pentium CPUs with intergrated graphics and it was good until you get Seton's or similar level of lagfest with 10 FPS.
And this athlon is basicly an overclocable pentium with better integrated GPU (afaik on b350 or better mobos that have agesa 6.0.0.0 you can OC CPU and GPU).... looks like a killer CPU for supcom.

i know some poeple that own really old laptops and no PC but want to play the game and they wonder how would cheapest supcom-capable PC look like.

Statistics: Posted by ZLO_RD — 02 Mar 2020, 15:58


]]>