Speed test of Intel 335 series SSD

S

sulfateion

Hi

I just bought a Intel 335 series SSD. I've used its migration tool to migrate windows 7 from my old 3.5 inch WD hard drive to SSD. After done, I reboot it and didn't find faster than before.
I check the speed of both ssd and WD HD and found that they are all 6Gb/s. Am I wrong somewhere?

Any suggestion?
 
R

Rod Speed

I just bought a Intel 335 series SSD. I've used its migration tool
to migrate windows 7 from my old 3.5 inch WD hard drive to SSD.
After done, I reboot it and didn't find faster than before. I check
the speed of both ssd and WD HD and found that they are all 6Gb/s.

What did you do to do the speed test ?
Am I wrong somewhere?

You sure you are actually running on the SSD ?
Any suggestion?

How does the speed of the system feel now ?
 
A

Arno

I just bought a Intel 335 series SSD. I've used its migration tool to migrate windows 7 from my old 3.5 inch WD hard drive to SSD. After done, I reboot it and didn't find faster than before.
I check the speed of both ssd and WD HD and found that they are all 6Gb/s. Am I wrong somewhere?
Any suggestion?

Yes. 6Gb/s = 600MB/s is the interface speed, not the media
access speed. Think of the speed a car tire is certified for
(interface speed), while the speed the car can go is the
media speed. Of course if media speed > interface speed,
things will get messy...

Arno
 
M

miso

Yes. 6Gb/s = 600MB/s is the interface speed, not the media
access speed. Think of the speed a car tire is certified for
(interface speed), while the speed the car can go is the
media speed. Of course if media speed > interface speed,
things will get messy...

Arno
I don't think your analogy is all that great. Presumably these SSDs have
some cache, so the cache can work at near the interface speed, while the
actual storage is slower.

Obviously this is a bigger deal with magnetic memory, since it is much
slower. You see this in benchmarks, where the write speed goes south
once the cache limit is reached.

If I could think of a better analogy, I would have posted it. I just
don't know a lot of "physical" things that behave like cached storage.
Well other than a grinder pump sewage system that some homes use. [Crap
goes quickly into a tank, but gets er um processed slowly before
pumping. ;-) ]
 
D

DevilsPGD

In the last episode of
I just bought a Intel 335 series SSD. I've used its migration tool to
migrate windows 7 from my old 3.5 inch WD hard drive to SSD. After done,
I reboot it and didn't find faster than before.
I check the speed of both ssd and WD HD and found that they are all
6Gb/s. Am I wrong somewhere?

To start off with, 6Gb/s is the speed of the interface, not the drives
themselves.
 
A

Arno

miso said:
On 11/6/2013 9:01 AM, Arno wrote:
I don't think your analogy is all that great. Presumably these SSDs have
some cache, so the cache can work at near the interface speed, while the
actual storage is slower.

Not really anymore. The cache would not be that much faster than
the medium, hence it has basically vanished in SSDs.

Arno
 
M

miso

Not really anymore. The cache would not be that much faster than
the medium, hence it has basically vanished in SSDs.

Arno

Do you have any hard numbers? I pulled this from the wiki:

Cache or buffer
------
A flash-based SSD typically uses a small amount of DRAM as a cache,
similar to the cache in hard disk drives. A directory of block placement
and wear leveling data is also kept in the cache while the drive is
operating. Data is not permanently stored in the cache.[39] One SSD
controller manufacturer, SandForce, does not use an external DRAM cache
on their designs, but still achieves very high performance. Eliminating
the external DRAM enables a smaller footprint for the other flash memory
components in order to build even smaller SSDs.[60]
--------

So some cache and some don't, if you believe the wiki. [You know how
that goes.]

The problem with googling SSD cache is you get mostly hits regarding
using SSDs themselves as a cache.
 
Y

Yousuf Khan

Hi

I just bought a Intel 335 series SSD. I've used its migration tool to migrate windows 7 from my old 3.5 inch WD hard drive to SSD. After done, I reboot it and didn't find faster than before.
I check the speed of both ssd and WD HD and found that they are all 6Gb/s. Am I wrong somewhere?

Any suggestion?

What's your speed testing program? Try ATTO, CrystalMark, and AS-SSD as
your benchmark programs.

How does the booting up time feel on the new drive?

As others have pointed out, the 6 Gb/s is just the interface speed, i.e.
just the maximum potential speed anything can possibly deliver on that
interface.

Yousuf Khan
 
A

Arno

miso said:
On 11/8/2013 12:20 PM, Arno wrote: [...]
Do you have any hard numbers? I pulled this from the wiki:
[...]

No. The thing is though that an SSD would need a huge
cache to have any visible effect and thet OS caches have gotten
good enough years back that disk caches do not matter. I think
they were primarily added to have a "larger" number than previous
generations.

Today, not even professional disks list any RAM cache. But
if you look at, e.g., the ~500MB/s (=6GBit/s with coding overhead)
that a fast SSD can deliver and store, that is already at the limit
of the interface and no amount of RAM would help. So I guess RAM
caches on SSDs do not make any sense and are not present anymore.

They did not make any sense on HDDs for quite some time either.

Arno
 
A

Arno

I just bought a Intel 335 series SSD. I've used its migration tool to migrate windows 7 from my old 3.5 inch WD hard drive to SSD. After done, I reboot it and didn't find faster than before.
I check the speed of both ssd and WD HD and found that they are all 6Gb/s. Am I wrong somewhere?
Any suggestion?

As to speed, maybe your bottle-neck is _not_ the disk?
SSDs do only make PCs faster if the disk in there
did slow it down before. All other things are exactly
the same speed with SSD and HDD.

Arno
 
M

miso

miso said:
On 11/8/2013 12:20 PM, Arno wrote: [...]
Do you have any hard numbers? I pulled this from the wiki:
[...]

No. The thing is though that an SSD would need a huge
cache to have any visible effect and thet OS caches have gotten
good enough years back that disk caches do not matter. I think
they were primarily added to have a "larger" number than previous
generations.

Today, not even professional disks list any RAM cache. But
if you look at, e.g., the ~500MB/s (=6GBit/s with coding overhead)
that a fast SSD can deliver and store, that is already at the limit
of the interface and no amount of RAM would help. So I guess RAM
caches on SSDs do not make any sense and are not present anymore.

They did not make any sense on HDDs for quite some time either.

Arno

I'm not sure what you call a professional disk, but all the magnetic
media (HDD) I have looked at have their cache posted on the box. Most
are 64Mbytes, some 128Mbytes. So they don't make sense and they still
use them?
 
R

Rod Speed

miso said:
Arno wrote
I'm not sure what you call a professional disk, but all the magnetic media
(HDD) I have looked at have their cache posted on the box. Most are
64Mbytes, some 128Mbytes. So they don't make sense and they still use
them?

He appeared to be talking about SSD professional disks, not magnet media
(HDD)s
 
A

Arno

miso said:
I'm not sure what you call a professional disk, but all the magnetic
media (HDD) I have looked at have their cache posted on the box. Most
are 64Mbytes, some 128Mbytes. So they don't make sense and they still
use them?

Professional = Enterpsise. Sorry.

The reason is that they want the number on the box. Customers
would not understand that the number is meaningless.

Arno
 
M

miso

Professional = Enterpsise. Sorry.

The reason is that they want the number on the box. Customers
would not understand that the number is meaningless.

Arno

But when you do benchmarks, you can see the effect of the cache on a
hard drive. The controller also uses the cache to put the data in
sequence if they do a read ahead.

What you really want to do is make sure your computer isn't doing a lot
of caching, especially with SSD. I just don't get this SSD caching
mania. I try to do as few writes to mine as possible. No virtual memory,
no file indexing, etc. There are lists of things to turn off when you
use a SSD.

It would be interesting to know just how deep the caching goes on a hard
drive. That is, you get 64Mbytes, but maybe it never needs more than 8
for your typical user.
 
A

Arno

But when you do benchmarks, you can see the effect of the cache on a
hard drive. The controller also uses the cache to put the data in
sequence if they do a read ahead.

No, you cannot. Unless you do bare disk benchmarks, which are
mostly meaningless for the performance in connection with an OS.

Track-buffers are not cache, BTW.
What you really want to do is make sure your computer isn't doing a lot
of caching, especially with SSD.

Exactly the opposite. The OS should do as much caching and buffering
as possible as RAM is always better, except for the problem that it
is volatile.
I just don't get this SSD caching
mania. I try to do as few writes to mine as possible. No virtual memory,
no file indexing, etc. There are lists of things to turn off when you
use a SSD.
It would be interesting to know just how deep the caching goes on a hard
drive. That is, you get 64Mbytes, but maybe it never needs more than 8
for your typical user.

You need almost none. Yjhe only thing beneficial it does is write
reordering with disk-geometry knowledge. The OS cannot do that as
modern disks lie abiut their geometry.

It gets fully used though. You can get into trouble because some
disks claim to have flushed the buffer contents to disk,
when they have not. This makes larger on-disk memory dangerous.
Linux kernel developers found this out by powering drives down
directly after the drive claimed to have flushed its buffers to
disk surface.

Arno
 
M

miso

No, you cannot. Unless you do bare disk benchmarks, which are
mostly meaningless for the performance in connection with an OS.

This I doubt. You see the benchmarks start to choke on large writes,
while the small ones are just cached to the drive.

Track-buffers are not cache, BTW.

When did I mention this?
Exactly the opposite. The OS should do as much caching and buffering
as possible as RAM is always better, except for the problem that it
is volatile.

Except you spend time caching to the sys ram, then the drive cachces it
again. Too many caches can be bad.
You need almost none. Yjhe only thing beneficial it does is write
reordering with disk-geometry knowledge. The OS cannot do that as
modern disks lie abiut their geometry.

Yet drives get better caches every year.
It gets fully used though. You can get into trouble because some
disks claim to have flushed the buffer contents to disk,
when they have not. This makes larger on-disk memory dangerous.
Linux kernel developers found this out by powering drives down
directly after the drive claimed to have flushed its buffers to
disk surface.

Arno
On linux, you just enter sync. Common when flashing OS on SSD, but not
limited to that task. However, you made my point. Just let the drive do
the caching.
 
A

Arno

miso said:
On linux, you just enter sync. Common when flashing OS on SSD, but not
limited to that task. However, you made my point. Just let the drive do
the caching.

The point is precisely that "sync" does not work reliably because
the drives are lying.

Arno
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top