OC3D Review: OCZ 64GB Core series SSD
- Thread starter w3bbo
- Start date
magicmerlin
New member
I had a regular Raptor before, but i sold it. And i thought i wouldnt notice a great difference when i had a regular drive, but i was very wrong on that point...That was a 150 GB Raptor...A question: In the test one SSD was tested, how much different results will i get when i do a raid-0 configuration with a good controllercard (ADAPTEC SATA RAID 1220SA SINGLE)? It must be like 30% better then?
Advanced Host Controller Interface allows Hot-Plugging and Native Commands Queuing as well as multithreaded access of the drive by applications. Enabling AHCI results in conflicts between the controller and the drive that are apparent as sluggish overall system performance. Benchmarks that actually work are those that are not geared towards quick and dirty assessment of a HDDs performance based on optimization of the test algorithms to meet the typical HDD architecture. Examples are PCMark Vantage, Winbench 99 2.0 "Drive Inspection Test" or ATTO.
Taken from Benchmarkreviews.com - if the reviewer didn't disable AHCI then seems this review is pretty much useless. There is no mention of him having done so in the intro to the article.
Taken from Benchmarkreviews.com - if the reviewer didn't disable AHCI then seems this review is pretty much useless. There is no mention of him having done so in the intro to the article.
name='FlyBy' said:
We run real file copy benchmarks, Game loading benchmarks, Windows startup benchmarks. All of which provide much more useful results than a "score" produced by applications such as PCMark.
name='FlyBy' said:
AHCI was not enabled. Drives were set to IDE Compatible mode - Thanks.
name='FlyBy' said:
As stated above, AHCI was disabled. I did initially think that AHCI had been enabled after running the benchmarks but a quick check confirmed it was still at its default [Disabled] setting. I therefore felt little need to inform the reader AHCI was disabled as most BIOS will have it disabled by default anyway.
With regards to the benchmarks, I was actually asked to run benchmarks that were more favourable to the OCZ Core drive but I questioned the validity of this as, by an OCZ employees own admission, the Core series does not appear to handle small files well, which naturally made me curious.
HD tune rightly or wrongly picks up on the 'issue' of the small files being written(which btw are not there with the original OCZ SSD) and therefore the results were included in the review for comparison purposes. The validity of the write results was however questioned, so to quote myself:
As you can see. I openly admit that the issue with HDtune write readings are not indicative of real world performance, at least not to the extent it shows. However this does show that the Core series drive DOES have issues where a stall appears when writing small files. Some benchmarks do not represent this - PCMark being one of them, ATTO being another, which if you read the review again you will see was used for clarification purposes. Not only that but we also tested the drive in 'real world scenarios'. Compression/decompression & file creation were also used, again emphasizing the drives capabilities. Synthetic benchmarks will always be a contentious issue and hence we also ran some 'everyday' tasks to see how it performs with the aim to present a balanced and informative review. Sometimes though the latest hardware is difficult to benchmark as there are simply no benchmarks written that take advantage of it and reflect its true performance!
To Quote OCZ:
Oh and I nearly forgot :
Taken from OCZ press release .Pdf file:
As quoted by you :
name='FlyBy' said:
If you feel the need to read (and thereby quote) a direct port of an OCZ press release then please feel free to do so but here at OC3D, we prefer to tell our readers how it is.
Nick at da Mill
New member
name='magicmerlin' said:
I know this is an old thread, but for what it's worth, my Adaptec 1220SA freezes when my OCZ CORE_SSD (32GB) is plugged into either port. I don't know if this is a generic problem with OCZ CORE_SSDs, or is specific to my card (OCZ suggests I RMA it to find out, which I will). Adaptec, claim "No SSD type drives have been tested with the 1220SA card, and currently the card only supports SATA model drives". I presume that by "SATA model drive" they mean "Sata HD". In any case, it supports a Samsung MCBQE32G5MPP-0VA (32GB) SSD without problems.
The OCZ CORE_SSD (32GB) works fine on a JMicron JMB36X controller, in tandem with a Samsung MCBQE32G5MPP-0VA (32GB), in a RAID0 configuration. In this configuration, I see a considerable improvement in load times, etc. over a pair of 150GB WD Raptors on the same controller and in the same configuration. Unfortunately, this controller does not write cache, nor does it overlap I/O across SATA devices, hence my desire to use another (better) controller.
Over the next week, after they arrive, I'm going to try a Promise FastTrack SX4 in place of the Adaptec 1220SA, and also a pair of Patriot Extreme Flash Warp (32GB) SSDs (which were on sale) on both controllers.
Finally, I have replaced a Samsung MCBQE32G5MPP-0VA (32GB) SSD with a OCZ CORE_SSD (64GB) on a two-disk bay HP Pavillion DV8219ea -- 32GB is just too small for Vista 32. Unfortunately, performance has gone to hell as a result. Vista hangs, with the disk activity LED pegged, from time to time. I've tried the "Windows Steady State" fix, but that makes no difference. Again, this may be a generic OCZ problem, or unique to my SSD.
I'll update as I find out more.
Nick at da Mill
New member
Patriot Extreme Flash Warp SSDs (32GB) also hang the Adaptec 1220SA. Waiting for the Promise Adapter to arrive. Hopefully before Christmas.
noobieocer
New member
still to expensive for em i stick to samsung f1 drives 
Nick at da Mill
New member
name='noobieocer' said:still to expensive for em i stick to samsung f1 drives
You're obviously not a speed freak. I'm using two SSDs to replace a pair of 150GB Raptors, both in a RAID0 configuration. Even on the JMicron controller, the SSDs (not yet the Patriots) have resulted in a 40% reduction in load time.
I got the Patriots on sale for £80.55 (ex VAT) each. Not very big, but 64GB is fine for a C: drive. My D: drive is 2TB (5-x-500 WD Caviars in a RAID5 config) on an Intel IH9.
Nick at da Mill
New member
Well the Promise controller (FastTrak S150 SX4) was a washout on my DFI LT-X48 MoBo - not recognized! Recognized under Vista X64 - but the 64 bit drivers aren't being recognized. Probably aren't signed in any case.
Recognized OK on an ASUS P5N-E SLI, but that already has an NVIDEA RAID controller that is working fine with a pair of Raptors in a RAID0 cionfig.
I'l try it in an old (RAID-less) server next.
I've now got an HP LSI SAS3041E SAS SATA RAID Controller on order. I'll wait to see how it fares on the DFI MoBo with a pair of Patriot SSDs, before upgrading my old server.
That leaves me with a spare Adaptec 1220SA with nowhere to go but eBay.
Recognized OK on an ASUS P5N-E SLI, but that already has an NVIDEA RAID controller that is working fine with a pair of Raptors in a RAID0 cionfig.
I'l try it in an old (RAID-less) server next.
I've now got an HP LSI SAS3041E SAS SATA RAID Controller on order. I'll wait to see how it fares on the DFI MoBo with a pair of Patriot SSDs, before upgrading my old server.
That leaves me with a spare Adaptec 1220SA with nowhere to go but eBay.
Nick at da Mill
New member
Attempting to use discrete RAID/SATA 0 Controllers with Patriot Extreme Flash Warp (32GB), or OCZ CORE_SSD (32GB), SSDs on my DFI LT X48 MoBo has been a complte bust!
None of the three SATA/RAID controllers were able to support the OCZ or Patriot SSDs!
The Adaptec 1220SA, hung when it had a Patriot Extreme Flash Warp (32GB) or OCZ CORE_SSD (32GB), but not a Samsung MCBQE32G5MPP-0VA (32GB), SSD plugged into either port.
The HP LSI SAS3041E SAS SATA RAID controller rejected the Patriot Extreme Flash Warp (32GB) SSDs ("Wrn Typ") for inclusion in a RAID array.
The Promise FastTrak S150 SX4 was not recognized by the MoBo.
So I bit the bullet and removed 2-x-500GB drives from the RAID5 array (6 drives -> 4 drives) on my integral Intel ICH9 RAID/SATA controller freeing two ports for an ICH9-based RAID0 array (32K stripe, with write-back caching) employing two Patriot Extreme Flash Warp (32GB) SSDs. This, except for losing a Terabyte of storage, was a great success!
Under Vista Ultimate X64, an ATTO benchmark (with a 256MB file) reported a 60MB/sec read & write transfer rate for 1KB blocks. The write rate got no better, but the read rate peaked at 180MB/sec on > 8K blocks. An HD Tune benchmark reported average read rates of 190 MB/sec with a min of 170MB/sec and a max of 195MB/sec, an average access time of 0.2 ms (vs. 10.2 ms for a 2-x-150GB Raptor RAID0 array), and an 886MB/sec burst rate.
I then tried the HP LSI SAS3041E SAS SATA RAID controller on an ASUS P5N-E SLI MoBo. With 4-x-36GB Raptors in a striped (64KB - no alternative) and mirrored RAID10 (IME in LSI terminology) array, the performance was so bad that I gave up. I then tried the 4-x-36GB Raptors in a RAID0 (64KB stripe - no alternative) array. Under Vista Ultimate X64, an ATTO benchmark (256MB file) reported very poor small block read and write performance (1 MB/sec with 1KB blocks), with write performance peaking at 240MS/sec on >= 64KB blocks, and read performance reaching 230MB/sec on >512KB blocks). HD Tune benchmarks reported an average read performance of 76MB/sec and a max burst rate of 91MB/sec. It is worth noting that this controller only supports write through (as opposed to write back) caching.
Finally, I tried the Promise FastTrak S150 SX on an MSI MS-7091 MoBo in a RAID5 (16KB stripe) array with three 500GB HDs. Under Win Server 2K, an ATTO benchmark (256MB and 32MB file) reported quite weird results. Small block read/write performance was good (16MB/sec for 1KB blocks), but read performance peaked at 57MB/sec on 16KB blocks and then dropped to 38MB/sec on larger blocks. My guess is that this controller cannot support parallel operations over multiple SATA ports. It is effectively operating SATA ports as if they were PATA ports. Write performance peaked at 50MB/sec on >64KB blocks. An HD Tune benchmark reported average read rates of 35MB/sec with a min of 28MB/sec and a max of 63MB/sec, an average access time of 13.2 ms, and a 90MB/sec burst rate.
These results were very poor when compared with a 4 HD Intel IH9-based RAID5 (32KB stripe) array on my DFI MoBo. Under Vista Ultimate X64, an ATTO benchmark (with 256MB file) reported a 53MB/sec write, and an 83MB/sec read, transfer rate for 1KB blocks. The write rate peaked at 112MB/sec > 4KB blocks, and the read rate peaked at 162MB/sec on 16K blocks, before dropping to 120MB/sec on > 64KB blocks. An HD Tune benchmark reported average read rates of 165 MB/sec with a min of 106MB/sec and a max of 207MB/sec, an average access time of 16.8 ms (vs. 10.2 ms for a 2-x-150GB Raptor RAID0 array), and an 871MB/sec burst rate.
This is probably much more data than anyone wants! The results are pretty clear. Low cost RAID/SATA controllers are a bust on most SSDs. They also don't perform that well with conventional HDs. Integral Intel and NVIDIA RAID/SATA controllers seem to be fine with SSDs and deliver much better RAID performance with both SSDs and HDs. An integral JMicron JMB36X-based RAID/SATA controller falls somewhere in between!
None of the three SATA/RAID controllers were able to support the OCZ or Patriot SSDs!
The Adaptec 1220SA, hung when it had a Patriot Extreme Flash Warp (32GB) or OCZ CORE_SSD (32GB), but not a Samsung MCBQE32G5MPP-0VA (32GB), SSD plugged into either port.
The HP LSI SAS3041E SAS SATA RAID controller rejected the Patriot Extreme Flash Warp (32GB) SSDs ("Wrn Typ") for inclusion in a RAID array.
The Promise FastTrak S150 SX4 was not recognized by the MoBo.
So I bit the bullet and removed 2-x-500GB drives from the RAID5 array (6 drives -> 4 drives) on my integral Intel ICH9 RAID/SATA controller freeing two ports for an ICH9-based RAID0 array (32K stripe, with write-back caching) employing two Patriot Extreme Flash Warp (32GB) SSDs. This, except for losing a Terabyte of storage, was a great success!
Under Vista Ultimate X64, an ATTO benchmark (with a 256MB file) reported a 60MB/sec read & write transfer rate for 1KB blocks. The write rate got no better, but the read rate peaked at 180MB/sec on > 8K blocks. An HD Tune benchmark reported average read rates of 190 MB/sec with a min of 170MB/sec and a max of 195MB/sec, an average access time of 0.2 ms (vs. 10.2 ms for a 2-x-150GB Raptor RAID0 array), and an 886MB/sec burst rate.
I then tried the HP LSI SAS3041E SAS SATA RAID controller on an ASUS P5N-E SLI MoBo. With 4-x-36GB Raptors in a striped (64KB - no alternative) and mirrored RAID10 (IME in LSI terminology) array, the performance was so bad that I gave up. I then tried the 4-x-36GB Raptors in a RAID0 (64KB stripe - no alternative) array. Under Vista Ultimate X64, an ATTO benchmark (256MB file) reported very poor small block read and write performance (1 MB/sec with 1KB blocks), with write performance peaking at 240MS/sec on >= 64KB blocks, and read performance reaching 230MB/sec on >512KB blocks). HD Tune benchmarks reported an average read performance of 76MB/sec and a max burst rate of 91MB/sec. It is worth noting that this controller only supports write through (as opposed to write back) caching.
Finally, I tried the Promise FastTrak S150 SX on an MSI MS-7091 MoBo in a RAID5 (16KB stripe) array with three 500GB HDs. Under Win Server 2K, an ATTO benchmark (256MB and 32MB file) reported quite weird results. Small block read/write performance was good (16MB/sec for 1KB blocks), but read performance peaked at 57MB/sec on 16KB blocks and then dropped to 38MB/sec on larger blocks. My guess is that this controller cannot support parallel operations over multiple SATA ports. It is effectively operating SATA ports as if they were PATA ports. Write performance peaked at 50MB/sec on >64KB blocks. An HD Tune benchmark reported average read rates of 35MB/sec with a min of 28MB/sec and a max of 63MB/sec, an average access time of 13.2 ms, and a 90MB/sec burst rate.
These results were very poor when compared with a 4 HD Intel IH9-based RAID5 (32KB stripe) array on my DFI MoBo. Under Vista Ultimate X64, an ATTO benchmark (with 256MB file) reported a 53MB/sec write, and an 83MB/sec read, transfer rate for 1KB blocks. The write rate peaked at 112MB/sec > 4KB blocks, and the read rate peaked at 162MB/sec on 16K blocks, before dropping to 120MB/sec on > 64KB blocks. An HD Tune benchmark reported average read rates of 165 MB/sec with a min of 106MB/sec and a max of 207MB/sec, an average access time of 16.8 ms (vs. 10.2 ms for a 2-x-150GB Raptor RAID0 array), and an 871MB/sec burst rate.
This is probably much more data than anyone wants! The results are pretty clear. Low cost RAID/SATA controllers are a bust on most SSDs. They also don't perform that well with conventional HDs. Integral Intel and NVIDIA RAID/SATA controllers seem to be fine with SSDs and deliver much better RAID performance with both SSDs and HDs. An integral JMicron JMB36X-based RAID/SATA controller falls somewhere in between!
