NCQ vs TCQ performance?
in [Linux Hardware]
|
From: Tom Forsmo on 6 May 2008 15:16 Arno Wagner wrote: > I did not calim that the Raptor was slow. What WD has claimed, however, > time and again, is that the Raptor is just as good as an SCSI disk. > It is not. I have never heard the WD marketing slogans. In any case, anyone knows that marketing is about exaggerating facts. Its always been that way, so if they have claimed its as fast?/good? as SCSI, its just marketing, otherwise we wouldn't need SCSI any more. But SCSI has other qualities which the Raptor or other SATAs does not have. E.g. SCSI scales with the load, which the Raptor does not. The Raptor only excels when the queue is larger than 16 commands, SCSI handles queues lower than that quite nicely in contrast. SCSI is also more stable, reliable and can handle top loads more stable than sata disks can. > It is, of course, faster than slower spinning (S)ATA > drives, unless you are interessted primarily in linear performance. What you are saying is that the only reason the raptor is faster is because of the spindle speed. which is not correct. Its is also faster because it uses TCQ instead of NCQ. But as stated, only for server loads. regards tom
From: Eric Gisin on 5 May 2008 13:41 TCQ was a disaster, it was built on top of a flawed attemp to make PATA multitask. I doubt anyone supports TCQ today, especially on SATA. Microsoft and Intel do NOT. NCQ performance varies from negative to positive compared to original IDE. You have to find some recent benchmarks, including both controllers and drives. "Tom Forsmo" <spam(a)nospam.net> wrote in message news:481f1b24$1(a)news.broadpark.no... > Hi > > In an article about the performance of sata raid controllers ( > StorageReview: > http://www.storagereview.com/articles/200406/20040625TCQ_1.html?page=0%2C5 > ) using NCQ or TCQ. it stated that TCQ is much faster than NCQ: > > "SATA TCQ and SATA RAID have the potential to deliver benefits to the > server market just as great as those of SCSI TCQ and SCSI RAID." > > So by using a SATA-1 TCQ RAID controller and WD Raptor disks one can get > close to SCSI performance. (Sata 2 does not actually give the same > performance according other tests I have read.) > > The article was written in 2004, and in four years much may have > happened with NCQ. So the question is, does anybody know if much has > happened with NCQ? or can I still trust the articles conclusions in > terms of its performance? >
From: Arno Wagner on 6 May 2008 16:44 In comp.sys.ibm.pc.hardware.storage Tom Forsmo <spam(a)nospam.net> wrote: > Arno Wagner wrote: >> I did not calim that the Raptor was slow. What WD has claimed, however, >> time and again, is that the Raptor is just as good as an SCSI disk. >> It is not. > I have never heard the WD marketing slogans. In any case, anyone knows > that marketing is about exaggerating facts. Its always been that way, so > if they have claimed its as fast?/good? as SCSI, its just marketing, > otherwise we wouldn't need SCSI any more. True. > But SCSI has other qualities which the Raptor or other SATAs does not > have. Also true. > E.g. SCSI scales with the load, which the Raptor does not. The > Raptor only excels when the queue is larger than 16 commands, SCSI > handles queues lower than that quite nicely in contrast. SCSI is also > more stable, reliable and can handle top loads more stable than sata > disks can. Agreed. SCSI also has multi-master capabilities and some other nice functional featuires. >> It is, of course, faster than slower spinning (S)ATA >> drives, unless you are interessted primarily in linear performance. > What you are saying is that the only reason the raptor is faster is > because of the spindle speed. which is not correct. No, that is not what I am saying. True, latency plays a role, but... > Its is also faster because it uses TCQ instead of NCQ. But as > stated, only for server loads. ... as you say. For mostly linear accesses TCQ/NCQ daoe not play a role and the lower bit density of the Raprors (and SCSI drives) slow them down to SATA comparable speeds. I expect by server loads you mean database loads, webserver loads and the like, i.e. multiple smaller reads from, potentially from different processes. Then the queuing model and latency are certainly or primary interest. Arno
From: Tom Forsmo on 6 May 2008 02:35 Eric Gisin wrote: > TCQ was a disaster, it was built on top of a flawed attemp to make PATA > multitask. > I doubt anyone supports TCQ today, especially on SATA. Microsoft and > Intel do NOT. Then I think you have been misinformed, some SATA disk support TCQ exceptionally well, such as the WD Raptor disks. Paired with a TCQ sata controller they outperform any NCQ disks on market. It actually comes close to SCSI performance, which no NCQ disks is able of achieving. Mind you this is with server type load. With single user load, the performance is somewhat lower than with NCQ. The reason for that is because TCQ is a more heavyweight protocol than NCQ, so you wont see a gain until the load is increased somewhat. You should read the entire article, it is quite interresting StorageReview: http://www.storagereview.com/articles/200406/20040625TCQ_1.html?page=0%2C5 > NCQ performance varies from negative to positive compared to original IDE. > You have to find some recent benchmarks, including both controllers and > drives. I have been looking around, and havent found much, neither benchmarks nor info about NCQ updates. regards tom
From: Arno Wagner on 6 May 2008 07:53 In comp.sys.ibm.pc.hardware.storage Tom Forsmo <spam(a)nospam.net> wrote: > Hi > In an article about the performance of sata raid controllers ( > StorageReview: > http://www.storagereview.com/articles/200406/20040625TCQ_1.html?page=0%2C5 > ) using NCQ or TCQ. it stated that TCQ is much faster than NCQ: > "SATA TCQ and SATA RAID have the potential to deliver benefits to the > server market just as great as those of SCSI TCQ and SCSI RAID." > So by using a SATA-1 TCQ RAID controller and WD Raptor disks one can get > close to SCSI performance. (Sata 2 does not actually give the same > performance according other tests I have read.) WD makes this claim for some time now, because they do not have SCSI disks. I think it is a marketing-lie. > The article was written in 2004, and in four years much may have > happened with NCQ. So the question is, does anybody know if much has > happened with NCQ? or can I still trust the articles conclusions in > terms of its performance? I think you copuld not trust it 2004 and you cannot trust it now. Arno
|
Next
|
Last
Pages: 1 2 Prev: Older Hardware, Older (smaller?) Distros to match Next: NCQ vs TCQ performance? |