| xmoonproductions.org https://www.xmoonproductions.org/ |
|
| 3.5 (+patch) Performance! https://www.xmoonproductions.org/viewtopic.php?f=16&t=3972 |
Page 3 of 3 |
| Author: | laissemoirire [ Sat Jul 18, 2015 1:19 pm ] |
| Post subject: | Re: 3.5 (+patch) Performance! |
b00marrows wrote: The cpu i was looking at, the ( i7-5820K ) has LESS l2 cashe than my current one @ 256KB per core ... but it dose have 15MB or L3 cashe... URGH i need more information... Oh, turns out My current CPU has 2MB of l2 cashe per core lol, the site i was using was wrong. Ah yes i should make a computer building thread next! im not sure how we could check for cpu cashe requirements tho... One thing I have not said, is that "the bigger the better" is not true for cache. To my knowledge, there is 2 reason for CPUs to have low cache size. - the price. - The time it takes to find data in cache. Even if cache acces is much more faster than RAM acces, cached values have to be searched (to be simple). If the requested memory data requested is not in cache (after searching), then the value is loaded from RAM. So, the bigger the cache is, the longer it takes to find if it has the requested data (or not). And if the cache is too big, it takes longer to find data in cache, than reading it directly from memory |
|
| Author: | b00marrows [ Sat Jul 18, 2015 5:04 pm ] |
| Post subject: | Re: 3.5 (+patch) Performance! |
laissemoirire wrote: b00marrows wrote: The cpu i was looking at, the ( i7-5820K ) has LESS l2 cashe than my current one @ 256KB per core ... but it dose have 15MB or L3 cashe... URGH i need more information... Oh, turns out My current CPU has 2MB of l2 cashe per core lol, the site i was using was wrong. Ah yes i should make a computer building thread next! im not sure how we could check for cpu cashe requirements tho... One thing I have not said, is that "the bigger the better" is not true for cache. To my knowledge, there is 2 reason for CPUs to have low cache size. - the price. - The time it takes to find data in cache. Even if cache acces is much more faster than RAM acces, cached values have to be searched (to be simple). If the requested memory data requested is not in cache (after searching), then the value is loaded from RAM. So, the bigger the cache is, the longer it takes to find if it has the requested data (or not). And if the cache is too big, it takes longer to find data in cache, than reading it directly from memory Wait... if it's not cashe size, but it's cache read speed... which is nothing to do the cpu... it's memory bus speed... which is ram/motherboard/chipset dependent...what? |
|
| Author: | laissemoirire [ Sat Jul 18, 2015 5:54 pm ] |
| Post subject: | Re: 3.5 (+patch) Performance! |
cache memory is integrated to the CPU, its speed is northbridge (and memory bus) independent. Cache memory speed is based on CPU clock rate, not on the FSB like any other PC component found a small picture that explain that : http://computer.howstuffworks.com/motherboard4.htm I don't really understand what your said, it's a question ? a question to yourself ? a statement ? |
|
| Author: | b00marrows [ Sat Jul 18, 2015 7:33 pm ] |
| Post subject: | Re: 3.5 (+patch) Performance! |
laissemoirire wrote: cache memory is integrated to the CPU, its speed is northbridge (and memory bus) independent. Cache memory speed is based on CPU clock rate, not on the FSB like any other PC component found a small picture that explain that : http://computer.howstuffworks.com/motherboard4.htm I don't really understand what your said, it's a question ? a question to yourself ? a statement ? Basically... My performance is limited by my ram/memory bus and/or chipset/northbridge speed and NOT my cpu. If this is true, it will be easier to know what i need to build on the desktop i am building. |
|
| Author: | laissemoirire [ Sat Jul 18, 2015 10:24 pm ] |
| Post subject: | Re: 3.5 (+patch) Performance! |
haaa ok if your 22FPS with 4 girls are related to memory load issue (what xpadmin supposes) changing your RAM to a low latency may improve the game. in a large majority of cases, low latency RAM increase overall performance by only 2-3% but theorically, you can have a 20% performance increase passing from 11-11-11-28 to 9-9-9-24 using the same memory speed after, I don't know. buy a new system with overpriced CPU, "new" DDR4, and tell us if it's better 
|
|
| Page 3 of 3 | All times are UTC |
| Powered by phpBB® Forum Software © phpBB Group http://www.phpbb.com/ |
|