If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#1
|
|||
|
|||
new ssd anticlimactic
put another one, ssd in
dos native transfers - no gain, if not pain, between ssdssd but it is a semi-intensive compression thingy I'm working with on sector-to-sector binary transfers ssd's are great, don't get me wrong, but at a certain point of efficiency, they're redundant. Once the OS and supportive applications layers are in NAND - it's only going so much farther. may as well call it hybrid scenario then, as storage to cost considerations mean a lot more when a $89 plattered 3T drive trumps (for most) a $500 1T SSD. Fortunately, I've got enough accumulated storage/data to where I can say getting around in that quagmire is a lot easier on a SSD than platters. (175G of storage still does take some time for eating up CPU cycles while churning over indices and whatnot. Transfers, figure I'm cutting time down somewhere between 1/2 and third of a conventional drive trxs.) As for my core OS, binary backups, one SSD pretty well covers about as much speed as the OS is going to give up;- as mentioned, DOS binary backups, well, it's in the header. They're a lot of fun, though. Real pain in the butt to get this one - had to fight it tooth and nail over getting the MBR settled, (Partition Magic fixed it - 1st fat/primary for valid fdisk/format to sys a: c: take, and last a boot arbitrator and that's pretty damn nervey for probably a 10yr old prg!), some trouble with Win XP puking over virtual memory (swapfile) assignments/cludge to get thru. Got that thru another "settings" angle, luckily. Haven't looked it over for what Win7 will think of my "new&improved sys" yet. This isn't a Samsung SSD unit, btw, like my other two. Samsungs are pretty much effortless in my experience. Still this new brand, it's working as expected with nothing major wrong - has a good reputation for a somewhat smaller adherence, 3yr warrantee &etc., so enough said about $20-25 average on 60G chunks of NAND. That trim crap - can't believe after all these years they don't break down and address some code directly to the controller. Was reading some crap that for the controller otherwise to kick in its "garbage routine" (independently of the non-TRIM OS) -- Log Out and let the computer sit for 10 hours. How about kiss my big hairy butt while I wait expressely for somebody to address code directly to the controller;- besides, it may be NTFS dependent for an OS TRIM request: screw TRIM for *nix and FAT altogether, sounds like. And, little surprise, I can't stand NTFS (except for what I'm stuck with using for - in case dealing in larger than FAT32 allowances of 4G filesizes). That stuff about writing 1's or 0's to clear the logic NAND gates, nahhh, don't like that either. Only thing I got going for me, is once settled in, I don't need to churn data, especially, and can just let it run 24/7, the way I like. Got a big'un in between the SDD units, besides, big plattered drive if I need to churn up some redundant read/write muck. |
#2
|
|||
|
|||
new ssd anticlimactic
Seems like you were writing about an Intel SSD, but you never said. Since you brought it up, does anyone know how well "trim" (automatic) works on the Crucial MX100 drives (512 GB, $219)? Bill Flasherly wrote: put another one, ssd in dos native transfers - no gain, if not pain, between ssdssd but it is a semi-intensive compression thingy I'm working with on sector-to-sector binary transfers ssd's are great, don't get me wrong, but at a certain point of efficiency, they're redundant. Once the OS and supportive applications layers are in NAND - it's only going so much farther. may as well call it hybrid scenario then, as storage to cost considerations mean a lot more when a $89 plattered 3T drive trumps (for most) a $500 1T SSD. Fortunately, I've got enough accumulated storage/data to where I can say getting around in that quagmire is a lot easier on a SSD than platters. (175G of storage still does take some time for eating up CPU cycles while churning over indices and whatnot. Transfers, figure I'm cutting time down somewhere between 1/2 and third of a conventional drive trxs.) As for my core OS, binary backups, one SSD pretty well covers about as much speed as the OS is going to give up;- as mentioned, DOS binary backups, well, it's in the header. They're a lot of fun, though. Real pain in the butt to get this one - had to fight it tooth and nail over getting the MBR settled, (Partition Magic fixed it - 1st fat/primary for valid fdisk/format to sys a: c: take, and last a boot arbitrator and that's pretty damn nervey for probably a 10yr old prg!), some trouble with Win XP puking over virtual memory (swapfile) assignments/cludge to get thru. Got that thru another "settings" angle, luckily. Haven't looked it over for what Win7 will think of my "new&improved sys" yet. This isn't a Samsung SSD unit, btw, like my other two. Samsungs are pretty much effortless in my experience. Still this new brand, it's working as expected with nothing major wrong - has a good reputation for a somewhat smaller adherence, 3yr warrantee &etc., so enough said about $20-25 average on 60G chunks of NAND. That trim crap - can't believe after all these years they don't break down and address some code directly to the controller. Was reading some crap that for the controller otherwise to kick in its "garbage routine" (independently of the non-TRIM OS) -- Log Out and let the computer sit for 10 hours. How about kiss my big hairy butt while I wait expressely for somebody to address code directly to the controller;- besides, it may be NTFS dependent for an OS TRIM request: screw TRIM for *nix and FAT altogether, sounds like. And, little surprise, I can't stand NTFS (except for what I'm stuck with using for - in case dealing in larger than FAT32 allowances of 4G filesizes). That stuff about writing 1's or 0's to clear the logic NAND gates, nahhh, don't like that either. Only thing I got going for me, is once settled in, I don't need to churn data, especially, and can just let it run 24/7, the way I like. Got a big'un in between the SDD units, besides, big plattered drive if I need to churn up some redundant read/write muck. |
#3
|
|||
|
|||
new ssd anticlimactic
On Sat, 23 Aug 2014 12:12:55 -0400, Bill
wrote: Seems like you were writing about an Intel SSD, but you never said. Since you brought it up, does anyone know how well "trim" (automatic) works on the Crucial MX100 drives (512 GB, $219)? So weird - not intel - in fact an $80 Crucial MX 256. Accepted convention for the automated Garbage Collection routine is just to leave it alone to do its thing. A few "magic" utilities intended to force the automated portion of GC are far and between, and generally only reinforce that aspect (of a, hm, not stupidity - but "faith" in technological intent for GC). |
#4
|
|||
|
|||
new ssd anticlimactic
On Sat, 23 Aug 2014 12:25:57 -0400, Flasherly
wrote: &btw - I filled mine twice - reformated and copied data to fill it twice to its capacity of 256G. During a course of mistakes (didn't catch a 64K segment format that should have been 4K segments) and establishing a primary partition both for validly to boot and take a boot arbitrator on the MBR (BIOS very first time up locked and rejected it, scary, and except for an old Partition Manager Pro 8.5, semi-scary, locked at a DMPI recognition every time for every all other MBR/format-sys/fdisk I used). Why I'd recommend a Samsung over that experience to anyone without tools and a persistance for hammering (on something not necessarily all about "Winderz"). |
#5
|
|||
|
|||
new ssd anticlimactic
Bill wrote:
Seems like you were writing about an Intel SSD, but you never said. Since you brought it up, does anyone know how well "trim" (automatic) works on the Crucial MX100 drives (512 GB, $219)? Bill This shows how they test for it. http://www.anandtech.com/show/4202/t...-510-review/13 "As expected, Intel's SSD 510 fully supports the ATA TRIM instruction. To gauge it's implementation I filled the 510 with data then ran a 4KB random write test (QD=32, 100% LBA space) for 20 minutes and measured performance using ATTO. I then TRIMed the entire drive by formatting it in Windows 7 and re-ran the ATTO test." The review of the MX100, on the other hand, doesn't even have the word TRIM in the article. The MX100 in the graph here, is a little slow on write, compared to some others. http://www.anandtech.com/show/8066/c...512gb-review/7 If you compare the Performance Consistency page in this article, to the MX100, you'll see they specifically test TRIM here, and the TRIM doesn't recover performance. It implies the MX100 doesn't have TRIM (as they didn't do a TRIM test). On the other hand, while the MX100 has "flat line graphs", I can't tell from the text description whether it recovered from the 4K write test or not. Certainly the IOP rate of the MX100 can drop to 20% of original value, as seen in one graph. But I don't understand exactly how those graphs relate to how the test for Performance Consistency works. http://www.anandtech.com/show/8294/intel-ssd-pro-2500/2 The review here claims the MX100 supports TRIM. Or rather, the reporting utility sees TRIM commands as being accepted. With TRIM, there's no way to know what the drive does with the information, which could just be ignored as it is considered a "hint" and not a "command". There is no useful information in this review, to comment on whether the TRIM is used or not. The lack of TRIM analysis on the Anandtech review, implies it isn't used for some reaslon. http://www.thessdreview.com/our-revi...ew-256-512-gb/ Considering the MX100 price, I think you're supposed to just buy it and use it :-) And not look too closely under the covers. I tried a search on the controller chip in the MX100, "88ss9189 and trim", and in another review, the command latency didn't bounce back when the pressure was off the drive. Unlike some competing drives. Try tracking down the 88ss9189 and see if there is more info on it. While radically different firmwares could be developed for it, most companies likely don't have designers with the skill set to do it (write a new firmware from scratch, not just turn some tuning knobs). Paul |
#6
|
|||
|
|||
new ssd anticlimactic
Paul wrote:
snip Thanks for your post. I read some of those articles about the MX100 earlier, but I didn't mind looking again! Bill |
#7
|
|||
|
|||
new ssd anticlimactic
On Sat, 23 Aug 2014 13:39:34 -0400, Paul wrote:
I tried a search on the controller chip in the MX100, "88ss9189 and trim", and in another review, the command latency didn't bounce back when the pressure was off the drive. Unlike some competing drives. Try tracking down the 88ss9189 and see if there is more info on it. While radically different firmwares could be developed for it, most companies likely don't have designers with the skill set to do it (write a new firmware from scratch, not just turn some tuning knobs). Here it is*, all else being equal to identical controllers (" "-BLD2) and some suggesting ADATA is just a rebadged Crucial. (For whatever else write performances might have to do with memory type implementation and subsequent rating considerations.) In any event - can't see a reason offhand for ADATA's TRIM implementation software not to work with the Marvel. Pretty slick looking actually (potentially for me, since I got nada else with this drive so far). Interesting, too, if it did actually help/improve with that latency delay;- not so sure it would, though, as the controller has been around long enough in various developmental forms for something of a stable regard, if not a "black box" implementation perhaps never quite properly translated from obscurer Far Eastern regions and dialects. * http://www.hardcoreware.net/adata-pr...pro-sp920-ssd/ |
Thread Tools | |
Display Modes | |
|
|