Traces of Various Games

[One_Hertz]

http://tweakers.net/benchdb/suite/33

This is really cool

They traced the I/O activity of different games and inputted the workload into the IPEAK benchmark so you can run their config files and get a good estimate of the performance of your storage array in various games.

This is REALLY cool . They also have a database of various setups to compare to.

[Griff805]

Horrible site... they wrote it all in the wrong language.

Thanks for the link though, looks like they did quite a bit of testing.

[One_Hertz]

Horrible site... they wrote it all in the wrong language.

Thanks for the link though, looks like they did quite a bit of testing.

Google translator works pretty well. This is how SSD reviews SHOULD be done, i.e. storagereview style.

[Spoiler]

Hmm, google translator doesn't seem to work for me too well with this site. Care to explain what's going on with each bench for each game. For example, what and how are they testing Counterstrike+Steam with their hard drive. Is this something I can test with my drive?

[One_Hertz]

Hmm, google translator doesn't seem to work for me too well with this site. Care to explain what's going on with each bench for each game. For example, what and how are they testing Counterstrike+Steam with their hard drive. Is this something I can test with my drive?

They are using IPEAK to measure IO activity during launching/loading/playing of various games and saving the results into a file. Using this file, other people can run the exact workload of those games in order to compare the gaming performance of storage setups to one another. This is even better than "real world" testing because real testing of loading times and such is highly influenced by CPU/RAM/Mobo etc. and reflects the performance of the whole system and not just the storage array.

The big problem with this is that IPEAK doesnt seem to work on 64bit OS. At least I haven't been able to run it yet.

[Derk]

I'll translate for you. The first part on the top of each benchmark says:

The (insert game name here)-benchmark is based on a trace of the diskactivity during the starting of (insert game name here), the loading and saving of a number of levels and closing the game. In the background eMule was active.

Lower there is a description of what they did.

Use RankDisk (part of Intel IPEAK storage Performance Toolkit) to test the workload on an unpartioned disk or array. The number of the wanted disk can be found in Windows Disk Management. rankDisk reports the average service time per I/O. This result can be converted to I/O's per second by dividing 1000ms by the service time (in milliseconds). Rankdisk can corrupt the data on the disk. That's why you should only test on an empty disk.

The Dutch text isn't written very well,that combined with me not understanding half leads to the translation not being perfect either. :P