Intel's SSD is faster than our PCs. Thus, a technology change is occurring.
Intel's USB IS Fastest. Expect Standards body to adopt it.
Both of these new additions are affecting Motherboard manufacturers who build AMD and Intel boards for user PCs. And, least we NOT forget AMD's recent CPU edition which is faster and less power-hungry than previous model.
This changes the landscape significantly and is already being addressed in Microsoft and Apple OS designs and their applications running in the OS.
FYI
Edited: New entries for USB and AMD advances.
New Intel SSD is faster than most every PC in 2015 Puppyland
-
gcmartin
New Intel SSD is faster than most every PC in 2015 Puppyland
Last edited by gcmartin on Wed 03 Jun 2015, 21:49, edited 4 times in total.
This is real!
This is the next step in SSD development.
It is in the first stages of being offered.
All major SSD manufactures are beginning to offer this new style of SSD.
The performance numbers are real.
Just how they will hook to the system, is still in final development process.
http://www.maximumpc.com/best-of-the-best-storage/
This is the next step in SSD development.
It is in the first stages of being offered.
All major SSD manufactures are beginning to offer this new style of SSD.
The performance numbers are real.
Just how they will hook to the system, is still in final development process.
http://www.maximumpc.com/best-of-the-best-storage/
The things they do not tell you, are usually the clue to solving the problem.
When I was a kid I wanted to be older.... This is not what I expected
YaPI(any iso installer)
When I was a kid I wanted to be older.... This is not what I expected
YaPI(any iso installer)
Numbers always help. Here is the speed
2400 MB/s Read, 1200 MB/s write
http://www.intel.com/content/www/us/en/ ... amily.html
The interface is is PCIe Gen 3 & 4
http://en.wikipedia.org/wiki/PCI_Express
which wikipedia gives 1,200 MB/s as the speed
Interestingly enough wikipedia's speed quote for PCIe gen 3 & 4 is slower than SATA revision 3.2 which is 1969 MB/s
http://en.wikipedia.org/wiki/Serial_ATA
The video is alleging that SATA can only deliver 550 MB/s,
https://youtu.be/hxpwmrmw6jI?t=21m21s
which is 1/4 of the speed that Wikipedia suggests that SATA can deliver.
2400 MB/s Read, 1200 MB/s write
http://www.intel.com/content/www/us/en/ ... amily.html
The interface is is PCIe Gen 3 & 4
http://en.wikipedia.org/wiki/PCI_Express
which wikipedia gives 1,200 MB/s as the speed
Interestingly enough wikipedia's speed quote for PCIe gen 3 & 4 is slower than SATA revision 3.2 which is 1969 MB/s
http://en.wikipedia.org/wiki/Serial_ATA
The video is alleging that SATA can only deliver 550 MB/s,
https://youtu.be/hxpwmrmw6jI?t=21m21s
which is 1/4 of the speed that Wikipedia suggests that SATA can deliver.
Sooooooo assuming those numbers actually are correct (which I find a little much, personally...) --
First off it's using PCIe which means that you can't boot from it except with very certain motherboards because PCIe stuff is traditionally nonbooting. That's a UEFI/BIOS issue.
Secondly, it's saturating the bus, at those speeds -- which means that when you access the drive, every other everything in your computer that is connected via PCIe has to sit on its hands until the transfer is complete and the bus is released. That is tremendously stupid, and the reason why should be tremendously obvious -- for the blissfully oblivious, that means nothing going to/from your graphics card(s) or discrete sound card or PCIe WiFi/LAN card. That is nasty! Not to mention that PCIe is not limited to your actual card slots -- there are for example LAN chipsets that rely on that even though they don't occupy a slot.
Further, you won't actually see those speeds. Those are 'ideal' speeds, they have to be -- which means that when you factor in things like errors and their correction, and processing overhead, and all that other inconvenient real-world crap... engineer's rule of thumb, divide by two.
All that still leaves one question -- what's the freaking point? Why do you actually need that kind of speed? or is it just bragging rights?
I'm computing on an HP T610 Plus "flexible" thin client. It's basically a small form factor desktop with an odd moniker... I'm more than happy with what I have. I cannot imagine needing or even wanting the kind of speed described here.
I also cannot imagine that kind of speed even being practical, particularly given what I laid out above.
My earlier comment stands.
First off it's using PCIe which means that you can't boot from it except with very certain motherboards because PCIe stuff is traditionally nonbooting. That's a UEFI/BIOS issue.
Secondly, it's saturating the bus, at those speeds -- which means that when you access the drive, every other everything in your computer that is connected via PCIe has to sit on its hands until the transfer is complete and the bus is released. That is tremendously stupid, and the reason why should be tremendously obvious -- for the blissfully oblivious, that means nothing going to/from your graphics card(s) or discrete sound card or PCIe WiFi/LAN card. That is nasty! Not to mention that PCIe is not limited to your actual card slots -- there are for example LAN chipsets that rely on that even though they don't occupy a slot.
Further, you won't actually see those speeds. Those are 'ideal' speeds, they have to be -- which means that when you factor in things like errors and their correction, and processing overhead, and all that other inconvenient real-world crap... engineer's rule of thumb, divide by two.
All that still leaves one question -- what's the freaking point? Why do you actually need that kind of speed? or is it just bragging rights?
I'm computing on an HP T610 Plus "flexible" thin client. It's basically a small form factor desktop with an odd moniker... I'm more than happy with what I have. I cannot imagine needing or even wanting the kind of speed described here.
I also cannot imagine that kind of speed even being practical, particularly given what I laid out above.
My earlier comment stands.
Maybe us the drive as a swap partition, in which case the transfers won't be as large and it will only be used when ram gets low.starhawk wrote:
Secondly, it's saturating the bus, at those speeds -- which means that when you access the drive, every other everything in your computer that is connected via PCIe has to sit on its hands until the transfer is complete and the bus is released. That is tremendously stupid, and the reason why should be tremendously obvious -- for the blissfully oblivious, that means nothing going to/from your graphics card(s) or discrete sound card or PCIe WiFi/LAN card. That is nasty! Not to mention that PCIe is not limited to your actual card slots -- there are for example LAN chipsets that rely on that even though they don't occupy a slot.
I thought about that -- that's actually the worst use case I can come up with off the cuff.
Remember that you can't use anything that's in swap -- it has to be pulled from swap into RAM before it's useful. So while you're avoiding RAM thrash by having swap, you're thrashing the heck out of your primary peripheral bus.
You can't have two devices send data along the same bus at once. That's inherent in bus design... really in the fundamentals of the function of basic electronics -- fundamentals which computers have to follow even though they aren't exactly 'basic' any more.
A bus is essentially a set of wires along which data travels, plus the protocol that governs how the data is to be sent and received. It's almost always at least somewhat parallel within a computer. (PCIe is more complicated -- it's parallel sets of serial transmission wires.) If you have a bus within which two devices are simultaneously trying to send data, then the two signals blend together (logic AND function) and the result is called a "bus collision". Nasty stuff.
So, while your PC's CPU is transferring data to and from this drive into RAM (or in the case of DMA, as the drive dumps to RAM directly) -- you cannot do anything else at all with any of your PCIe bus. That is electrically and electronically not permitted.
So... if you're torrenting another Linux distro (let's keep this legal!), whenever Transmission (or uTorrent or ...) needs to send data to the SSD, it hogs the bus for that transfer. Conversely, if anything else is using that bus, Transmission (etc) has to wait until the bus is free. This isn't as much an issue with say your graphics card (which is hardware-incapable of saturating an AGP bus -- there was no real reason for PCIe in the first place) as it is with this SSD.
If you are transferring multiple gigs of data, or really files of any meaningful size, you're going to have a massive performance penalty with the entire rest of your computer. Rather than having a hard drive that is itself slow, requiring the computer to wait on it -- this SSD slows down the computer around it so that you cannot take full advantage of your system while the SSD is transferring data.
That's actually kind of creepy, to be perfectly honest.
Remember that you can't use anything that's in swap -- it has to be pulled from swap into RAM before it's useful. So while you're avoiding RAM thrash by having swap, you're thrashing the heck out of your primary peripheral bus.
You can't have two devices send data along the same bus at once. That's inherent in bus design... really in the fundamentals of the function of basic electronics -- fundamentals which computers have to follow even though they aren't exactly 'basic' any more.
A bus is essentially a set of wires along which data travels, plus the protocol that governs how the data is to be sent and received. It's almost always at least somewhat parallel within a computer. (PCIe is more complicated -- it's parallel sets of serial transmission wires.) If you have a bus within which two devices are simultaneously trying to send data, then the two signals blend together (logic AND function) and the result is called a "bus collision". Nasty stuff.
So, while your PC's CPU is transferring data to and from this drive into RAM (or in the case of DMA, as the drive dumps to RAM directly) -- you cannot do anything else at all with any of your PCIe bus. That is electrically and electronically not permitted.
So... if you're torrenting another Linux distro (let's keep this legal!), whenever Transmission (or uTorrent or ...) needs to send data to the SSD, it hogs the bus for that transfer. Conversely, if anything else is using that bus, Transmission (etc) has to wait until the bus is free. This isn't as much an issue with say your graphics card (which is hardware-incapable of saturating an AGP bus -- there was no real reason for PCIe in the first place) as it is with this SSD.
If you are transferring multiple gigs of data, or really files of any meaningful size, you're going to have a massive performance penalty with the entire rest of your computer. Rather than having a hard drive that is itself slow, requiring the computer to wait on it -- this SSD slows down the computer around it so that you cannot take full advantage of your system while the SSD is transferring data.
That's actually kind of creepy, to be perfectly honest.
-
bark_bark_bark
- Posts: 1885
- Joined: Tue 05 Jun 2012, 12:17
- Location: Wisconsin USA
-
gcmartin
You might want to listen to the interview. Pay attention to "lanes" used in Intels. And, try to understand what is shared in the architected performance of the CPU (today's), the buses (yes that's plural), the MB changes made (that is promised also by the MB vendors for some of the older boards), and how this integrations supports the workloads in OS operations (yes this means Linux too). This is simply great integrations for use and sustainability over time.
Today, I personally cannot afford neither Samsung's/Toshiba's 3D or Intel's offering, but, I do understand where this is headed and how it will be beneficial in Home use (when I can afford them in future purchases).
Hope this is helpful. Over and out.
Today, I personally cannot afford neither Samsung's/Toshiba's 3D or Intel's offering, but, I do understand where this is headed and how it will be beneficial in Home use (when I can afford them in future purchases).
Hope this is helpful. Over and out.
-
gcmartin
How do you get that?
Maybe its better to just accept the 5year warranty with 70GB/day of write cycles that is mentioned as a measure of whether this would meet anyone's requirements. If not, then this is NOT the device for you. In fact, this SSD has one of the highest in industry numbers for its kind.
But, listen, again to the interview. Maybe there is something practical available in understanding of what is changing and how that is beneficial to desktop home use.
And, we must not forget that traditional backup is still strongly recommended in use of PC(s) and any data system(s). This fact will never change; until we are totally immersed in the Cloud where that responsibility rest with those providers.
Maybe its better to just accept the 5year warranty with 70GB/day of write cycles that is mentioned as a measure of whether this would meet anyone's requirements. If not, then this is NOT the device for you. In fact, this SSD has one of the highest in industry numbers for its kind.
But, listen, again to the interview. Maybe there is something practical available in understanding of what is changing and how that is beneficial to desktop home use.
And, we must not forget that traditional backup is still strongly recommended in use of PC(s) and any data system(s). This fact will never change; until we are totally immersed in the Cloud where that responsibility rest with those providers.
http://www.techradar.com/us/reviews/pc- ... 008/review
Gen 3 PCIe slot and watch it go.
There are certain caveats though - it needs to be at least a x4 lane slot, and it needs to be connected directly to the CPU rather than the PCH. It also needs the full bandwidth of a PCIe 3.0 slot to give the SSD 750 a chance to really stretch its NVMe legs.
Plug it into the PCH and you'll only get around half the performance.
It's also only available on Z97 and X99 boards for now.
This PCIe version is fully bootable - once our Asus Maximus VII Hero board had its latest BIOS in place, it instantly saw the SSD 750, gave us the option to install our OS and boot away. Setting up the OS was a doddle too, though Microsoft's NVMe driver needed replacing with Intel's latest driver to get top speed.
Benchmarks
Peak sequential read performance
ATTO - MB/s: higher is better
Intel SSD 750 1.2TB - 2671
Samsung XP941 512GB - 1060
Samsung 850 Pro 512GB - 561
Peak sequential write performance
ATTO - MB/s: higher is better
Intel SSD 750 1.2TB - 1322
Samsung XP941 512GB - 997
Samsung 850 Pro 512GB - 532
Average sequential read performance
AS SSD - MB/s: higher is better
Intel SSD 750 1.2TB - 2136
Samsung XP941 512GB - 1055
Samsung 850 Pro 512GB - 506
Average sequential write performance
AS SSD - MB/s: higher is better
Intel SSD 750 1.2TB - 1175
Samsung XP941 512GB - 885
Samsung 850 Pro 512GB - 500
4k random read performance
AS SSD - MB/s: higher is better
Intel SSD 750 1.2TB - 38
Samsung XP941 512GB - 23
Samsung 850 Pro 512GB - 37
4k random write performance
AS SSD - MB/s: higher is better
Intel SSD 750 1.2TB - 200
Samsung XP941 512GB - 49
Samsung 850 Pro 512GB - 107
Real-world compression performance
5GB folder - Seconds: quicker is better
Intel SSD 750 1.2TB - 69
Samsung XP941 512GB - 73
Samsung 850 Pro 512GB - 73
Real-world transfer performance
30GB folder - Seconds: quicker is better
Intel SSD 750 1.2TB - 120
Samsung XP941 512GB - 132
Samsung 850 Pro 512GB - 169
Our testing platform is a stock-clocked Intel Core i7-4770K on an Asus RoG Maximus VII Hero Z97 motherboard with 8GB Corsair Dominator DDR3 at 1,600MHz.
Gen 3 PCIe slot and watch it go.
There are certain caveats though - it needs to be at least a x4 lane slot, and it needs to be connected directly to the CPU rather than the PCH. It also needs the full bandwidth of a PCIe 3.0 slot to give the SSD 750 a chance to really stretch its NVMe legs.
Plug it into the PCH and you'll only get around half the performance.
It's also only available on Z97 and X99 boards for now.
This PCIe version is fully bootable - once our Asus Maximus VII Hero board had its latest BIOS in place, it instantly saw the SSD 750, gave us the option to install our OS and boot away. Setting up the OS was a doddle too, though Microsoft's NVMe driver needed replacing with Intel's latest driver to get top speed.
Benchmarks
Peak sequential read performance
ATTO - MB/s: higher is better
Intel SSD 750 1.2TB - 2671
Samsung XP941 512GB - 1060
Samsung 850 Pro 512GB - 561
Peak sequential write performance
ATTO - MB/s: higher is better
Intel SSD 750 1.2TB - 1322
Samsung XP941 512GB - 997
Samsung 850 Pro 512GB - 532
Average sequential read performance
AS SSD - MB/s: higher is better
Intel SSD 750 1.2TB - 2136
Samsung XP941 512GB - 1055
Samsung 850 Pro 512GB - 506
Average sequential write performance
AS SSD - MB/s: higher is better
Intel SSD 750 1.2TB - 1175
Samsung XP941 512GB - 885
Samsung 850 Pro 512GB - 500
4k random read performance
AS SSD - MB/s: higher is better
Intel SSD 750 1.2TB - 38
Samsung XP941 512GB - 23
Samsung 850 Pro 512GB - 37
4k random write performance
AS SSD - MB/s: higher is better
Intel SSD 750 1.2TB - 200
Samsung XP941 512GB - 49
Samsung 850 Pro 512GB - 107
Real-world compression performance
5GB folder - Seconds: quicker is better
Intel SSD 750 1.2TB - 69
Samsung XP941 512GB - 73
Samsung 850 Pro 512GB - 73
Real-world transfer performance
30GB folder - Seconds: quicker is better
Intel SSD 750 1.2TB - 120
Samsung XP941 512GB - 132
Samsung 850 Pro 512GB - 169
Our testing platform is a stock-clocked Intel Core i7-4770K on an Asus RoG Maximus VII Hero Z97 motherboard with 8GB Corsair Dominator DDR3 at 1,600MHz.
The things they do not tell you, are usually the clue to solving the problem.
When I was a kid I wanted to be older.... This is not what I expected
YaPI(any iso installer)
When I was a kid I wanted to be older.... This is not what I expected
YaPI(any iso installer)
I assume this amazing new technology (which I assume it is) is going to be heavily bogged down once again with severely bloated and inefficient software.
If only the quality of software matched that of the hardware computers would really fly
Machines have multiplied performance several times over the last 20 years yet seem to perform the same tasks at a slightly faster pace...anyone looking into that paradox? A slight but not overstated exaggeration..
mike
If only the quality of software matched that of the hardware computers would really fly
Machines have multiplied performance several times over the last 20 years yet seem to perform the same tasks at a slightly faster pace...anyone looking into that paradox? A slight but not overstated exaggeration..
mike
I blame javascript and flash and any other tool used to make highly bloated web pages. Of course bloated operating systems are also a problem. Fortunately, we don't have to use a bloated OS. We can use Puppy instead.mikeb wrote:I assume this amazing new technology (which I assume it is) is going to be heavily bogged down once again with severely bloated and inefficient software.
If only the quality of software matched that of the hardware computers would really fly
Machines have multiplied performance several times over the last 20 years yet seem to perform the same tasks at a slightly faster pace...anyone looking into that paradox? A slight but not overstated exaggeration..
mike
Well i was thinking of the systems Intel will be aiming at which will be predominately Windows...more a general observation rather than puppy based which is one example of a more rational operating system.
Operating systems are for running programs/apps but that seems to have been forgotten. They seem to be now an end in itself.
You also have ridiculous pdf readers and printer drivers all taking hundreds of megs...just plain software lunacy.
And antivirus must get the award for the most bloated useless software ever...
So my point... why drown hardware technology in crap software?
mike
Operating systems are for running programs/apps but that seems to have been forgotten. They seem to be now an end in itself.
You also have ridiculous pdf readers and printer drivers all taking hundreds of megs...just plain software lunacy.
And antivirus must get the award for the most bloated useless software ever...
So my point... why drown hardware technology in crap software?
mike
you just reminded me of when experimenting with file sharing I had made a boo boo with a script and was suddenly faced with rox filer in front of me showing folders disappearing one by one before my eyes and nothing I did seem to stop it..indeed it was a few seconds before i realised what was happening. One partition emptied in 10 seconds flat.All this just means you can lose data faster.
It gets you right in the stomach
Happy ending was a 99% recovery due to backups and file recovery..I learned a lot that day..anyone want a debugfs tutorial
Plus i learned the difference between rm -rf and rmdir and hate to see the former used....
Ok that was an aside... please continue
mike