To be fair, the xp941 is a first generation Samsung SATA Express AHCI ssd from 2013. Samsung is set to release the 5th generation PCIe ssd's within 30 days. Performance has scaled tremendously over the last 5 years
Don't buy from OWC. Buy the cheapest 10600R sticks. CPU is 3 channels. 1, 2 or 4 will degrade performance.
SATA 2 and 3 won't really make a huge difference. Only with very big files. Not worth if you have a SATA SSD already.
Yes, I realised that some latest high end consumer SSD can deliver 100MB/s on the 4k random read test. Which will be noticeable faster than an average SATA SSD can do. But the cost is also noticeable higher (for the same capacity)
View attachment 775419
Yes, I realised that some latest high end consumer SSD can deliver 100MB/s on the 4k random read test. Which will be noticeable faster than an average SATA SSD can do. But the cost is also noticeable higher (for the same capacity)
View attachment 775419
While 1,500 MBP/sec reads / writes seems impressive it's not really relevant to the work that a lot of people do. It represents the sequential speed and not the random speed. IMO the strength of SSDs, for most users, is in their random performance and not the sequential performance. While > 1GB/sec speeds look impressive it's not the type of disk access which benefits most users. Random performance is where most people benefit and, in many cases, it has not reached SATA-II speeds. Therefore moving to a PCIe AHCI or NVMe is unlikely to provide any tangible results for many people. Blindly suggesting PCIe based solutions may lead to disappointment.It's also good to remember that while a SATA II SSD delivers the fastest boot time in the cMP, that's about the only performance lead it delivers. Operations in mail, photos, imovie, etc that are disk I/O intensive will be much slower than SATA III which is much slower than a modern PCIe SSD.
When considering any SATA III ssd, consumer oriented chipsets whether asmedia or marvell, introduces boot time delays and slower 4k performance compared to any Intel SATA II or III controller. Slightly higher file transfer, by a couple hundred mb/s is the only benefit.
As I have been stating for years, running MacOS from a PCIe SSD, delivers a class of performance not attainable from SATA II or III. Without a special controller, a 2009+ cMP can achieve 1500 mb/s reads and writes.
While 1,500 MBP/sec reads / writes seems impressive it's not really relevant to the work that a lot of people do. It represents the sequential speed and not the random speed. IMO the strength of SSDs, for most users, is in their random performance and not the sequential performance. While > 1GB/sec speeds look impressive it's not the type of disk access which benefits most users. Random performance is where most people benefit and, in many cases, it has not reached SATA-II speeds. Therefore moving to a PCIe AHCI or NVMe is unlikely to provide any tangible results for many people. Blindly suggesting PCIe based solutions may lead to disappointment.
While 1,500 MBP/sec reads / writes seems impressive it's not really relevant to the work that a lot of people do. It represents the sequential speed and not the random speed. IMO the strength of SSDs, for most users, is in their random performance and not the sequential performance. While > 1GB/sec speeds look impressive it's not the type of disk access which benefits most users. Random performance is where most people benefit and, in many cases, it has not reached SATA-II speeds. Therefore moving to a PCIe AHCI or NVMe is unlikely to provide any tangible results for many people. Blindly suggesting PCIe based solutions may lead to disappointment.
The system I'm responding to your post on has both a 256GB M.2 AHCI SSD as well as a 512GB M.2 NVMe. It's an HP Z620 with a Z-Turbo drive (M.2 AHCI) boot disk and a Z-Turbo G2 drive (M.2 NVMe) used for my virtual systems (the Z-Turbo G2 will not work as a boot drive in the Z620 system). The G2 drive replaced an SATA 850 EVO drive. I speak from experience.Are you kidding? I've got more that one of each kind of drive in my system and nothing, and I mean NOTHING SATAII comes close to the performance of PCIe SSD/NVMe SSD in pretty much everything. PCIe SSD (AKA: SSUBX) isn't far behind an NVMe as a daily driver, but there is a difference. Both in cost, and performance in daily use. SATAII isn't even on the same planet.
You must be hearing a different sound than reality.I also have some swap land in DC for sale if you are interested. It sounds like you are still using a SATA II SSD along with some traditional spinning hard disks.
Then please provide us with your random read/write benchmarks instead of the sequential ones.This is no blind recommendation, my vision is still great, Random read/writing speeds are also substantially faster. Your web browser will load faster, although you probably will not see any performance increase posting here. As I have said since 2012, PCIe SSD's deliver an enhanced cMP experience that a SATA SSD cannot match.
The system I'm responding to your post on has both a 256GB M.2 AHCI SSD as well as a 512GB M.2 NVMe. It's an HP Z620 with a Z-Turbo drive (M.2 AHCI) boot disk and a Z-Turbo G2 drive (M.2 NVMe) used for my virtual systems (the Z-Turbo G2 will not work as a boot drive in the Z620 system). The G2 drive replaced an SATA 850 EVO drive. I speak from experience.
What would be the point? I already stated the M.2 NVMe drive replaced a SATA 850 EVO drive with no discernible difference.Great, now put in an SATAII drive, and tell me it's just as fast as your current configuration.
There's a significant difference between SATA-III limitations and PCIe NVMe yet the NVMe drive does not provide any perceptible difference in "day to day" usage. This tells me that SATA is not a limiting factor.The system you're posting from is much newer and faster to begin with. You already have PCIe 3.0, and SATAIII capabilities.
I agree, the PCIe based solutions, in sequential throughput, greatly exceed that of SATA. Yet, in "day to day" usage the PCIe based solutions offer no perceptible benefit over SATA SSD solutions.Coming from an SATAII connection, it's not even in the same ballpark as a fast M.2 AHCI SSD or an NVMe SSD.
The system I'm responding to your post on has both a 256GB M.2 AHCI SSD as well as a 512GB M.2 NVMe. It's an HP Z620 with a Z-Turbo drive (M.2 AHCI) boot disk and a Z-Turbo G2 drive (M.2 NVMe) used for my virtual systems (the Z-Turbo G2 will not work as a boot drive in the Z620 system). The G2 drive replaced an SATA 850 EVO drive. I speak from experience.
[doublepost=1534026605][/doublepost]
You must be hearing a different sound than reality.
Then please provide us with your random read/write benchmarks instead of the sequential ones.
Or you could, you know, support your position here where you've made your claims.This conversation has drifted too far 'OFF-TOPIC'. I've posted countless benchmarks from 2009 and 2010 cMP's from Quickbench showing 4k performance on a variety of SSD's, Hard Disks, and PCIe media. While I could post #'s from a x58 Gigabyte Hackintosh to compare with your HP, that probably belongs over at tonymacx86.
Good luck with your HP
You did not achieve 1,500MB/sec with a SATA-III drive. It was that metric which I responded to, not some SATA-II versus SATA-III metric.My point was, you were using an SATAIII connection to your SATAIII SSD.
I was coming from an SATAII connection. There is a HUGE perceptible difference in day to day usage for me. It's not even close. I've felt a performance bump with each iteration of drive upgrade I've done.
Disagree all you'd like, it won't change the facts.I disagree with you, as I have my own experience and you have yours.
LOL! Can't support your position so you try and bury it.We disagree, let's put a fork in it and call it day.
Or you could, you know, support your position here where you've made your claims.
[doublepost=1534030611][/doublepost]
You did not achieve 1,500MB/sec with a SATA-III drive. It was that metric which I responded to, not some SATA-II versus SATA-III metric.
Disagree all you'd like, it won't change the facts.
LOL! Can't support your position so you try and bury it.
Not sure what MP3's / MP4's, Photoshop, or PDF files have to do with booting a system. If it's your argument the sequential throughput of PCIe based SSD's can outperform those of SATA based SSD's you'll get no argument from me.How many photos are 4k in size? MP3's/mp4's? How about photoshop files. PDF's. Not many data files are 4k in size anymore. We're not running on an OS based on the 8-bit z-80..
If you have a specific post or posts you'd like me to review then please point me to them. I am not going to read through a thread in search of...well, I don't know what you're asking me to find.
What benchmarks are you expecting from me?For comparison to your benchmarks(still waiting):
Agreed, which is why your statements are being challenged.Many have opinions, but you can never tell which end it's coming from.
What facts?Plain and simple, I run on facts. There is too much FAKE NEWS out there.
Not sure what MP3's / MP4's, Photoshop, or PDF files have to do with booting a system. If it's your argument the sequential throughput of PCIe based SSD's can outperform those of SATA based SSD's you'll get no argument from me.
However booting a system doesn't involve a lot of large file sequential reads. Instead it primarily consists of random reads. Therefore quoting sequential read metrics is useless.
If you have a specific post or posts you'd like me to review then please point me to them. I am not going to read through a thread in search of...well, I don't know what you're asking me to find.
What benchmarks are you expecting from me?
Agreed, which is why your statements are being challenged.
What facts?
Oh grow up!So.... What's your point beyond disagreement? I could color code the benchmarks I just posted, showing the random read performance variances between devices, but the crayons needed won't transfer through the screen.