|
Two 950's in R0 |
Post Reply 
|
Page 12> |
| Author | |
DooRules 
Newbie 
Joined: 05 Nov 2015 Location: Newfoundland Status: Offline Points: 122 |
 Post Options
 Thanks(0)
 Quote Reply
 Topic: Two 950's in R0Posted: 12 Dec 2015 at 8:37am |
|
My best score to date. Didn't hurt that my chip was at 4.9
  |
|
 |
|
|
parsec
Moderator Group 
Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
Post Options
Thanks(0)
Quote Reply
Posted: 12 Dec 2015 at 2:17pm |
|
Nice, nice, nice!
 You didn't lose much 4K random Read speed either compared to a single 950, less than 10%, my usual RAID 0 nitpick.  Being a RAID 0 user veteran, you know how the RAID 0 scaling with SATA SSDs usually goes, particularly with Sequential performance. We don't see that here with the Seq Read speed unfortunately. The write performance looks really good, much better scaling in the sequential and both 4K results, I'd love to see the AS SSD IOPs for that test. A few questions for you:
 |
|
|
|
|
|
|
DooRules
Newbie
Joined: 05 Nov 2015 Location: Newfoundland Status: Offline Points: 122 |
Post Options
Thanks(0)
Quote Reply
Posted: 12 Dec 2015 at 5:24pm |
|
Hello Parsec.
This is not an OS volume, still just playing with the drives.  - I am on UEFI 2.1, and for this run I was on 14.6.0.1029. But they are indistinguishable from 14.8 as far as I can see. - The stripe size was 128k. I have tried all of them at this point from 4k to 128k. IMHO these drives do better as the stripe size increases. Would love to see the 256k, 512k and 1MB option for an array to try. - Tried the ram at lower frequencies, 2133 and at XMP 3200, no discernible difference to me. What really does make a difference is chip frequencies. Higher clocked the chip, faster the bench, 1 to 1 ratio on that. - I did break one array. Not quite sure how i did it. I do suspect it had something to do with my GPU benching though. Once I get to the limit of the cards I end up with system freezes, hard restarts and the like.  Gee, who would have guessed, lol.- All caching options are available yes through the IRST CP for the array, and device manager. - I am somewhat baffled as to why the read scales so poorly in comparison the writes. But that gives me something to work on so all good. another couple days of playing around and then I will put OS on one drive and storage for the second. Ditch all other drives at that point.ps.. I could use a bios that allows for more than 1.5 Vc
Edited by DooRules - 12 Dec 2015 at 5:26pm |
|
|
|
|
DooRules
Newbie
Joined: 05 Nov 2015 Location: Newfoundland Status: Offline Points: 122 |
Post Options
Thanks(0)
Quote Reply
Posted: 13 Dec 2015 at 8:09pm |
|
Used a 60 gb file to see transfer speed between the two 950's , maintained the same speed for full transfer
 |
|
|
|
|
parsec
Moderator Group
Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
Post Options
Thanks(0)
Quote Reply
Posted: 14 Dec 2015 at 11:12pm |
|
That's great! Thanks for your answers.
A bit odd how it started out slow and then increased like crazy, usually I see the opposite, always dropping off at the end or midway. I've seen some 950 Pro users say they "feel" no difference between their 950 and good SATA SSDs. My Win 10 cold startup after the POST beep seems almost instant, while my SanDisk Extreme Pros take about three seconds. Driver and software installation is super fast, and things like copying my Windows installation driver and software collections of ~1.5GB in one folder, to a backup SSD (zero HDDs in my PCs  ), is subjectively instant, I don't see the Windows file copy dialog box appear at all.I think we are seeing the limitations of the Windows file system code, which was designed for HDDs, being at least part of the limitation of perceived and actual performance of all high end SSDs. Edited by parsec - 14 Dec 2015 at 11:12pm |
|
|
|
|
|
|
|
Xaltar
Moderator Group
Joined: 16 May 2015 Location: Europe Status: Offline Points: 24653 |
Post Options
Thanks(0)
Quote Reply
Posted: 14 Dec 2015 at 11:55pm |
That would make a lot of sense. I have noticed that my indexing/search times seem inordinately slower than most other tasks on my SSDs. Both of which functions utilize a very outdated sequential search method that, while effective on mechanical drives, fail to take advantage of the way an SSD accesses data. I would love to see MS focus more on updating their software to accommodate current tech and less on trying to force everyone to conform to how they want us to use their OS
 |
|
|
|
|
parsec
Moderator Group
Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
Post Options
Thanks(0)
Quote Reply
Posted: 15 Dec 2015 at 2:15pm |
|
I recall a forum post I read, where the guy was unhappy with the need for the "stupid TRIM" command for SSDs.
TRIM is a work around for the current way something in the Windows file system (NTFS?) was designed when only HDDs, floppy disks, and apparently USB flash drives existed. For HDDs, when files are deleted, all that is done is the space that was used by the files (LBAs) are put on the available list. An HDD happily overwrites itself if told to do so. SSDs, no. This unseen and unknown quirk of NAND storage that came after magnetic drives, which are the basis for computer storage IO software, is only one of the things that is not optimal for flash storage. It seems to me that the magnetic storage user base in businesses and governments alone will prevent any change to HDD oriented storage system software for quite a while. Meanwhile, NAND flash storage is evolving (PCIe SSDs, NVMe, M.2), and an apparent successor from Intel/Micron is on the horizon. Microsocialist's one OS for every device and person will impede progress in a different way. You cannot satisfy all of the people all of the time with one "thing", but if the number of people that are satisfied with that one thing is large enough, the rest of us will be ignored. The truly first portable PC (smart phones) has become that one thing IMO. IMO, we are lucky that we (still) have what we have now, and that it works well, if not optimally. Not that I disagree with you Xaltar at all, I don't. I just hope we don't lose it soon. |
|
|
|
|
|
|
|
DooRules
Newbie
Joined: 05 Nov 2015 Location: Newfoundland Status: Offline Points: 122 |
Post Options
Thanks(0)
Quote Reply
Posted: 15 Dec 2015 at 11:41pm |
 Throw in a cinebench run  Edited by DooRules - 16 Dec 2015 at 12:23am |
|
|
|
|
Xaltar
Moderator Group
Joined: 16 May 2015 Location: Europe Status: Offline Points: 24653 |
Post Options
Thanks(0)
Quote Reply
Posted: 16 Dec 2015 at 1:43am |
|
Those are some great scores DooRules. Makes me wish I had the budget to pick up a pair of 950s myself. Just out of curiosity, what scores are you getting on your 6700K with the CPUZ bench tool? Since the latest update it seems to be scaling better on multi core tests.
I doubt we will have to worry about it for a long time yet, not with so many people still utilizing mechanical drives for storage. It would be nice to see a new NAND specific file system that takes better advantage of the technology as an option. I know next to nothing about programming and development but I would imagine it would probably be a nightmare to make it play nice with NTFS and FAT32. That said, given how fast my system feels using a regular SSD and with so much faster available now it may just be a moot point by the time it were actually ready to roll out. The nerd in me just wants another excuse to geek out.
 |
|
|
|
|
DooRules
Newbie
Joined: 05 Nov 2015 Location: Newfoundland Status: Offline Points: 122 |
Post Options
Thanks(0)
Quote Reply
Posted: 16 Dec 2015 at 2:37am |
|
First time I have run that bench...
 |
|
|
|
|
Post Reply
|
Page 12> |
Tweet
|
| Forum Jump | Forum Permissions You cannot post new topics in this forum You cannot reply to topics in this forum You cannot delete your posts in this forum You cannot edit your posts in this forum You cannot create polls in this forum You cannot vote in polls in this forum |
Topic Options
parsec wrote: