User(s) browsing this thread: 1 Guest(s)

Post Reply 
 
Thread Rating:
  • 1 Vote(s) - 5 Average
  • 1
  • 2
  • 3
  • 4
  • 5
[blog] MMU mini-series (2/4)
Author Message
Blyss Sarania Offline
Imaginary Goddess
*******

Posts: 14.272
Joined: Sep 2013
Location: 私の夢の中
Reputation: 326
Post: #31
RE: [blog] MMU mini-series (2/4)
(02-22-2015 09:45 PM)willkuer Wrote:  That's again quite funny. Normally I would say that if you replace all cached stuff by volatile stuff it will still work but will be slower while enabling caching can help increasing performance... You are saying more or less the contrary.
That the whole system is slower if cache is enabled sounds at least possible to me if there is some overhead in implementing the cache. That disabling the cache breaks games on the other hand sounds completely strange.

I haven't ever seen but one game that needs it personally, and I forget what it was. As far as the performance hit, I was going by what it says in the UI "(Slower)"

[Image: sig.png]
AMD FX 6300 @ 4.51Ghz | GTX 670 (4GB) | 8GB RAM | 240GB SSD | 1 TB HDD
Tutorial: How to import GameFAQS saves into PCSX2

PCSX2 General Troubleshooting FAQ
(This post was last modified: 02-22-2015 11:31 PM by Blyss Sarania.)
02-22-2015 11:29 PM
Find all posts by this user Quote this message in a reply

Sponsored links

ssakash Offline
Bug squad
*****

Posts: 6.791
Joined: May 2014
Location: 127.0.0.1
Reputation: 100
Post: #32
RE: [blog] MMU mini-series (2/4)
(02-22-2015 09:45 PM)willkuer Wrote:  That's again quite funny. Normally I would say that if you replace all cached stuff by volatile stuff it will still work but will be slower while enabling caching can help increasing performance... You are saying more or less the contrary.
That the whole system is slower if cache is enabled sounds at least possible to me if there is some overhead in implementing the cache. That disabling the cache breaks games on the other hand sounds completely strange.
Did you read Gregory's article ?, he has mentioned multiple times that, cache is only useful when the data has temporal locality. If, it's gonna be accessed only one time then, it will only prove to be a disadvantage.

INTEL I7 4770|gtx 750ti|16gb ram|1tb hdd|160gb Intel ssd

Low budget PC building guide for PCSX2
02-23-2015 04:20 AM
Visit this user's website Find all posts by this user Quote this message in a reply
gregory Offline
Linux PCSX2 coder
******

Posts: 3.261
Joined: May 2010
Location: Grenoble, France
Reputation: 92
Post: #33
RE: [blog] MMU mini-series (2/4)
Don't mix your PC cache and the emulated (or not) EE cache.

As I said, cache contains a copy of a small part of the RAM. PCSX2 doesn't bother with cache emulation because you can get all data directly from the emulated RAM. From a PS2 point of view, it means that all access are uncached. However your PC will do cached access all the times. It is not an issue because PC have big cache. My L3 cache is 20MB whereas the full mem of the PS2 is only 32MB...

Now there are some exception to the rule RAM==CACHE. For example, a volatile memory not declared as volatile. A typical SW bug! It would be possible that data in RAM is different of the cache. A game might rely on this bug/feature, so EE-Cache will need to be emulated. Every time you do a PS2 memory access, you need to update cache (lots of computing). Rules of thumb: more hardware to emulate => slower speed.
02-23-2015 11:06 AM
Find all posts by this user Quote this message in a reply
willkuer Online
Posting Freak
****

Posts: 3.194
Joined: Oct 2010
Location:
Reputation: 87
Post: #34
RE: [blog] MMU mini-series (2/4)
(02-23-2015 04:20 AM)ssakash Wrote:  Did you read Gregory's article ?
You have some problems with courtesy, recently. I don't know what I or others did to you but I apologize for that. I hope you can talk to me in a polite manner in the future.

(02-23-2015 11:06 AM)gregory Wrote:  Rules of thumb: more hardware to emulate => slower speed.

This is clear. That's why I said
Quote:That the whole system is slower if cache is enabled sounds at least possible to me if there is some overhead in implementing the cache.


On the other hand this:
(02-23-2015 11:06 AM)gregory Wrote:  A game might rely on this bug/feature, so EE-Cache will need to be emulated.
is really strange.

If you rely on this method/bug/feature you get unexpected behavior, or? You will never know when the next time the cache is updated. So the feature that accesses a volatile but cached address will yield different results at different times. (sounds a bit like a randomizer function). Whatever the reason for the developers is to use a volatile address in cache...
Again thank you for answering!
02-23-2015 11:29 AM
Find all posts by this user Quote this message in a reply
gregory Offline
Linux PCSX2 coder
******

Posts: 3.261
Joined: May 2010
Location: Grenoble, France
Reputation: 92
Post: #35
RE: [blog] MMU mini-series (2/4)
Quote:is really strange.

If you rely on this method/bug/feature you get unexpected behavior, or? You will never know when the next time the cache is updated. So the feature that accesses a volatile but cached address will yield different results at different times. (sounds a bit like a randomizer function). Whatever the reason for the developers is to use a volatile address in cache..
Let's imagine a bug. You don't know the bug exists but you will do test on the real console. There are 2 possibilities
1/ application is bad => debug => fix the bug
2/ application is working like a charm (even with a bug). It is really possible, some hardware delay could make it work.

The emulator has a different set of delay, the previously hidden bug will appears... (note: It could also be a bug in the emulator but I wanted to highligth that you have no guarantee on anything).
02-24-2015 10:35 AM
Find all posts by this user Quote this message in a reply
willkuer Online
Posting Freak
****

Posts: 3.194
Joined: Oct 2010
Location:
Reputation: 87
Post: #36
RE: [blog] MMU mini-series (2/4)
Evil (evil smiley)
(This post was last modified: 06-14-2015 10:09 PM by willkuer.)
06-14-2015 10:09 PM
Find all posts by this user Quote this message in a reply
Post Reply 





Current time: 07-31-2015, 01:01 AM