• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Intel's First Client Optane Product is a Cache SSD

btarunr

Editor & Senior Moderator
Staff member
Joined
Oct 9, 2007
Messages
46,283 (7.69/day)
Location
Hyderabad, India
System Name RBMK-1000
Processor AMD Ryzen 7 5700G
Motherboard ASUS ROG Strix B450-E Gaming
Cooling DeepCool Gammax L240 V2
Memory 2x 8GB G.Skill Sniper X
Video Card(s) Palit GeForce RTX 2080 SUPER GameRock
Storage Western Digital Black NVMe 512GB
Display(s) BenQ 1440p 60 Hz 27-inch
Case Corsair Carbide 100R
Audio Device(s) ASUS SupremeFX S1220A
Power Supply Cooler Master MWE Gold 650W
Mouse ASUS ROG Strix Impact
Keyboard Gamdias Hermes E2
Software Windows 11 Pro
Intel's first consumer (client) SSD based on its revolutionary new 3D Xpoint memory is an Optane branded cache SSD that improves the performance of slower local storage, such as hard drives, or even slower NAND flash based SSDs. On machines with larger hard drives, Intel claims that a 3D Xpoint based cache SSD could halve booting times, improve overall system performance by 28 percent, and lower game level load times by up to 65 percent. As a cache-SSD, it's also designed to be affordable, and that's because it's local storage is 16 GB or 32 GB.

The target consumer is one that which is transitioning from hard drives to SSDs, and is happy with a noticeable performance boost, as long as they don't lose the immense capacities of their HDDs. It also targets gamers with SSDs that are running out of space for multiple >50 GB games, so they could start installing some of those games on their larger/slower HDDs and get reasonably improved performance. As with all SSD caching technologies from Intel in the past, such as the ReadyBoost and Smart Response, Optane cache SSDs juggle "hot data" (frequently accessed data) in and out of their user-space from the host storage. On the software side of things, Intel Rapid Storage Technology 15.5 and later handles the caching tasks.



The Optane cache SSD is a single-sided M.2-2280 drive with PCI-Express 3.0 x2 host interface, and takes advantage of the NVMe protocol. The drives offer sequential speeds of up to 1200 MB/s reads, with up to 280 MB/s writes, and 4K random access performance of up to 300,000 IOPS reads, with up to 70,000 IOPS writes. What Optane brings to the table is a gargantuan command queue depth, and an endurance rating of 100 GB writes per day (which is huge for a 16 or 32 GB drive, given that it's over 6 DWPD for the 16 GB variant.



View at TechPowerUp Main Site
 
Last edited:
Joined
May 21, 2008
Messages
4,113 (0.71/day)
Location
Iowa, USA
System Name THE CUBE 2.0
Processor Intel i5 13600k
Motherboard MSI MPG Z690 EDGE DDR4
Cooling Phanteks PH-TC14PE BK 2x T30-120 Fan mod mount
Memory G.Skill TridentZ 3200 MT/s C15 32GB 2x16GB
Video Card(s) Gigabyte Aorus 1080 Ti 11GB OC: Core 2GHz, Mem 5.7GHz
Storage WD SN770 250GB / 3x WD SN850X 2TB / Toshiba X300 4TB / 2x RAID1 Toshiba P300 3TB
Display(s) Samsung 49" Odyssey OLED G95SC 240Hz 5120 x 1440
Case "THE CUBE" Custom built, pure Red Alder wood
Audio Device(s) Beyerdynamic DT 880
Power Supply Corsair RM1000X
Mouse Logitech G700
Keyboard Logitech G910
Software Windows 11 Pro


Source:
 
Joined
Oct 27, 2009
Messages
1,129 (0.21/day)
Location
Republic of Texas
System Name [H]arbringer
Processor 4x 61XX ES @3.5Ghz (48cores)
Motherboard SM GL
Cooling 3x xspc rx360, rx240, 4x DT G34 snipers, D5 pump.
Memory 16x gskill DDR3 1600 cas6 2gb
Video Card(s) blah bigadv folder no gfx needed
Storage 32GB Sammy SSD
Display(s) headless
Case Xigmatek Elysium (whats left of it)
Audio Device(s) yawn
Power Supply Antec 1200w HCP
Software Ubuntu 10.10
Benchmark Scores http://valid.canardpc.com/show_oc.php?id=1780855 http://www.hwbot.org/submission/2158678 http://ww

Now we should compare to a traditional SSD then a NVME drive.

IF it competes with a traditional SSD and costs $40... It has a budget play.
Buy a $100 4tb drive, and a $40 16gb cache buffer and have ssd ish performance?
 
Joined
May 21, 2008
Messages
4,113 (0.71/day)
Location
Iowa, USA
System Name THE CUBE 2.0
Processor Intel i5 13600k
Motherboard MSI MPG Z690 EDGE DDR4
Cooling Phanteks PH-TC14PE BK 2x T30-120 Fan mod mount
Memory G.Skill TridentZ 3200 MT/s C15 32GB 2x16GB
Video Card(s) Gigabyte Aorus 1080 Ti 11GB OC: Core 2GHz, Mem 5.7GHz
Storage WD SN770 250GB / 3x WD SN850X 2TB / Toshiba X300 4TB / 2x RAID1 Toshiba P300 3TB
Display(s) Samsung 49" Odyssey OLED G95SC 240Hz 5120 x 1440
Case "THE CUBE" Custom built, pure Red Alder wood
Audio Device(s) Beyerdynamic DT 880
Power Supply Corsair RM1000X
Mouse Logitech G700
Keyboard Logitech G910
Software Windows 11 Pro
Now we should compare to a traditional SSD then a NVME drive.

IF it competes with a traditional SSD and costs $40... It has a budget play.
Buy a $100 4tb drive, and a $40 16gb cache buffer and have ssd ish performance?

My guess is when comparing to a SSD the difference wasn't much, so Intel didn't have them test it or it's coming in a future video.

Boot 39.74
On my system I got Excel 1.19Sec. Word 1Sec.
I don't have the others to test.
 
Last edited:
Joined
Dec 10, 2015
Messages
545 (0.18/day)
Location
Here
System Name Skypas
Processor Intel Core i7-6700
Motherboard Asus H170 Pro Gaming
Cooling Cooler Master Hyper 212X Turbo
Memory Corsair Vengeance LPX 16GB
Video Card(s) MSI GTX 1060 Gaming X 6GB
Storage Corsair Neutron GTX 120GB + WD Blue 1TB
Display(s) LG 22EA63V
Case Corsair Carbide 400Q
Power Supply Seasonic SS-460FL2 w/ Deepcool XFan 120
Mouse Logitech B100
Keyboard Corsair Vengeance K70
Software Windows 10 Pro (to be replaced by 2025)
For ultrabook it's just a waste of space, for budget build you better get an SSD for that kind of money

Maybe you can use this to cache the games that you installed in the HDD and shorten it's loading time, but I'm not sure if it's worth it
 
Joined
Oct 7, 2013
Messages
344 (0.09/day)
My guess is when comparing to a SSD the difference wasn't much, so Intel didn't have them test it or it's coming in a future video.

Boot 39.74
On my system I got Excel 1.19Sec. Word 1Sec.
I don't have the others to test.

excel and word are mostly single thread limited in terms of loading speed. Most of windows is single thread due to crappy coding. Hence why for my main rig i run a 4.8GHz 6700K.

you'll notice that cache speed is very different. If i open work close open excel and open word again. there is no cache. If i open and close and open word again the second time is cached. Windows need to be threaded and cache better...hell even let uses choose what to cache given the amount of RAM they have. I would have a slew of things preload in my RAM on boot if i could.

this is very easy to tell how crappy windows caching is and how much faster cached program loads when you play around with calc, paint, word, excel. Cache appears to dumb as soon as another program opens.

If you have NVMe vs RAMDisk and watch CPU in taskmanager and TS you will quickly see how nearly everything in windows is single thread limited.
 
Joined
Oct 2, 2004
Messages
13,791 (1.94/day)
Buy cheapest smallest SSD you can buy. Buy PrimoCache. Pair together to a HDD. Cheap SSD performance. There is just no way this Optane thing will be affordable. Not even at such tiny capacities. Cheap 128GB SSD cache will be just better as it'll cache WAY more stuff. And with SSD's, random access is all it matters, not massive sequential speeds. And difference between normal SSD and Optane just isn't that big in real world scenarios.
 
Joined
Oct 7, 2013
Messages
344 (0.09/day)
Buy cheapest smallest SSD you can buy. Buy PrimoCache. Pair together to a HDD. Cheap SSD performance. There is just no way this Optane thing will be affordable. Not even at such tiny capacities. Cheap 128GB SSD cache will be just better as it'll cache WAY more stuff. And with SSD's, random access is all it matters, not massive sequential speeds. And difference between normal SSD and Optane just isn't that big in real world scenarios.

optane excels at QD1...cheap SSDs are garbage as QD1 compared to XPoint (not counting this crap cache). Latency is also a huge deal.

960 PRO has done a good job of getting the latency down but optane is still better from what little leaks we have seen. Waiting for reviews to really tell.

QD 1 is all that matters...not "random ____"....not high QD and not sequential.

But the difference in NVMe NAND and XPoint won't really show until we get rid of PCIe and put this stuff in a direct channel to the CPU via memory lanes or some new port or better on the die.

Once you get XPoint closer to the CPU the QD1 IOPs will explode because PCH and PCIe have huge latency overhead do to being so far from the CPU.

http://www.anandtech.com/show/10754/samsung-960-pro-ssd-review/4

"expensive" SATA SSDS have 8-30 times the latency. Also consistency is crap.
 
Joined
Oct 2, 2004
Messages
13,791 (1.94/day)
In theory, that translates to massive differences. But in REAL situations, it means absolutely NOTHING. I've had that Sandisk ReadyCache which was based on USB thumbdrive controller/NAND. Then I bought Samsung SM951 M.2 AHCI with more IOPS and 1.5GB/s sequential read. The difference was minimal, mostly only at write when populating cache...
 
Joined
Oct 7, 2013
Messages
344 (0.09/day)
In theory, that translates to massive differences. But in REAL situations, it means absolutely NOTHING. I've had that Sandisk ReadyCache which was based on USB thumbdrive controller/NAND. Then I bought Samsung SM951 M.2 AHCI with more IOPS and 1.5GB/s sequential read. The difference was minimal, mostly only at write when populating cache...
you dont know what your talking about or what to look for. This is clear.

QD1 affect everything from AV/AM, explorer, loading, reading writing, encryption, everything.

A real XPoint SSD should show 3-7x better QD1 and thats very noticeable. I noticed from from 480GB Extreme pro to 950PRO easily. I also hammer my drives and i also use encryption and other stuff. Consistency is also critical here. Sandisk extreme pro was king of latency and consistency and mix load but the 950PRO rekts it and is very noticeable.

If XPoint was 2 bucks a GB i would be buying a TB drive in a heart beat because the difference is very noticeable especially in day to day activity. Again as i keep spamming the web with this.....go read IBMs 1980s study on rapid response systems.

Key here is response times and APM (actions per minute). There are many activities that are inherently limited by QD1 if you actually compare the best SATA to 950PRO and watch CPU and disk usage.

This is from 1982....yet people 35 years later still fail to understand basic computers and human brian fucntion. This is why i build my PC around QD1 and Signle thread since everything is coded like shit. AV/AM can't even use multiple thread or hgh QD because people can't code.

http://jlelliotton.blogspot.com/p/the-economic-value-of-rapid-response.html
 
Joined
Oct 2, 2004
Messages
13,791 (1.94/day)
Dude, I've used hybrid SSD/HDD storage when most of you were wanking on garbage SSD boot drives. Yeah, I KNOW what I'm talking about.
 
Joined
Oct 7, 2013
Messages
344 (0.09/day)
Dude, I've used hybrid SSD/HDD storage when most of you were wanking on garbage SSD boot drives. Yeah, I KNOW what I'm talking about.
yea...thats convincing. I have used SSDs, HDDs, RAIDs, NVMe for various tasks on very fast single thread PCs and there is huge differences when you use programs like PS, Chrome, Opera, Explorer, VeraCrypt, ripping files, 7zip, and so on.

Hell an uncrypted 480GB Extreme Pro could not download and install a game on Steam at 22MBps without pausing to play catch up because of the whole downloading, unpacking, and installing going on. The download was 4:1 in extracted to downloaded size on steam.

950PRO and single thread are still the bottleneck on malwarebytes (because of shitty coding) (well they did just release a multithreaded client so i haven't really analyzed the new one yet.

The difference from 950PRO to Extreme Pro in encryption is horrifyingly good.

The difference with extracting and installing on the same drive is the difference between night and day over those two drives.

I could keep going but i won't waste my time.

So yes you clearly are full of crap and have no clue what you are saying.
 
Joined
Oct 2, 2004
Messages
13,791 (1.94/day)
And you've just floated from upgrading garbage HDD performance with cheap SSD caching into triple digit prices of high end SSDs... You just make zero sense.

The jump from sole HDD to HDD with SSD cache is gigantic, almost identical to the level of replacing HDD with SSD entirely. The jump from crappy SSD to top end SSD is far less significant and certainly not justifiable with the price point. Even more less when that person still has HDD. Which can only mean they don't have the money. Otherwise they'd be on SSD only.
 
Joined
Oct 7, 2013
Messages
344 (0.09/day)
And you've just floated from upgrading garbage HDD performance with cheap SSD caching into triple digit prices of high end SSDs... You just make zero sense.

The jump from sole HDD to HDD with SSD cache is gigantic, almost identical to the level of replacing HDD with SSD entirely. The jump from crappy SSD to top end SSD is far less significant and certainly not justifiable with the price point. Even more less when that person still has HDD. Which can only mean they don't have the money. Otherwise they'd be on SSD only.

you claimed there is no difference between XPoint and NVMe in real world usage, which is factual wrong. I proved there are numerous places where there are drastic differences and plenty of room to gain in real world usage...your the one who makes zero sense lol.

In theory, that translates to massive differences. But in REAL situations, it means absolutely NOTHING.


I never defended this XPoint cache in any of my posts. I also said its a stupid garbage product unless in a real SSD form factor and wont be epic til we give it direct access to the CPU.

EDIT: additionally, you claim its not justified given the price point (crappy SSD to top of the line). Again, I will refer you to the IBM study from 1982 since after 35 years people are still idiots and fail to understand how humans work.

Their study is a prime reason why i remove all transitions from my phone and from windows and use a fast single thread CPU.
 
Last edited:
Joined
Oct 2, 2004
Messages
13,791 (1.94/day)
Because there isn't any worth making god damn drama about. And stuff that damage control BS wherever you want it, because you damn well know even the difference between average SSD and top of the line NVMe in REAL day to day use is almost non existent. If you say otherwise you're just lying to yourself. And consequently to others...
 

Frick

Fishfaced Nincompoop
Joined
Feb 27, 2006
Messages
18,914 (2.86/day)
Location
Piteå
System Name Black MC in Tokyo
Processor Ryzen 5 5600
Motherboard Asrock B450M-HDV
Cooling Be Quiet! Pure Rock 2
Memory 2 x 16GB Kingston Fury 3400mhz
Video Card(s) XFX 6950XT Speedster MERC 319
Storage Kingston A400 240GB | WD Black SN750 2TB |WD Blue 1TB x 2 | Toshiba P300 2TB | Seagate Expansion 8TB
Display(s) Samsung U32J590U 4K + BenQ GL2450HT 1080p
Case Fractal Design Define R4
Audio Device(s) Line6 UX1 + some headphones, Nektar SE61 keyboard
Power Supply Corsair RM850x v3
Mouse Logitech G602
Keyboard Cherry MX Board 1.0 TKL Brown
VR HMD Acer Mixed Reality Headset
Software Windows 10 Pro
Benchmark Scores Rimworld 4K ready!
A real XPoint SSD should show 3-7x better QD1 and thats very noticeable. I noticed from from 480GB Extreme pro to 950PRO easily. I also hammer my drives and i also use encryption and other stuff. Consistency is also critical here. Sandisk extreme pro was king of latency and consistency and mix load but the 950PRO rekts it and is very noticeable.

You talk about different use scenarious. Of course encryption will, as you say, hammer the drives. You are an edge case, RejZoR is closer to the avarage. You are both right, which should be obvious.
 
Joined
Oct 7, 2013
Messages
344 (0.09/day)
Because there isn't any worth making god damn drama about. And stuff that damage control BS wherever you want it, because you damn well know even the difference between average SSD and top of the line NVMe in REAL day to day use is almost non existent. If you say otherwise you're just lying to yourself. And consequently to others...

Again you don't understand human brain and how delays hurt productivity.

Go read that link and educate yourself. The windows fade animation is 120-150ms IIRC and thats is counter productive. No fade is about 30-50ms IIRC. The difference in significant especially when you open and close a window 100s of times a day. IBM study stopped at 300ms but it shows you got going from 600ms to 300ms saves 3 seconds in total time on the task at hand. Every task is difference but the concept on how the human brain works is universal. This is relevant to every aspect of computer use from proper UI design....which doesn't exist because idiots design this stuff for looks....thanks apple.....to the hardware in the PC and how it is coded. The difference from using my Ultrabook with a 3GHz SKL and Sammy EVO vs my desktop with a 4.8 GHz SKL and 950 PRO is huge.

You are truly ignorant on this topic and you should take awhile and read the study and learn. Pulling your head out of your ass is a good thing to do.



Additionally, consistency is a big issue from cheap SATA to best SATA to 950 PRO. I have posted extensively on hardforum about this. I just posted many day to day activity above about the best SATA vs 950 PRO where I notice much better user experience in numerous tasks. This ranges from loading chrome/opera to loading photos and moving files.

Again I would gladly ditch my 950 PRO for a XPoint if it was 2 dollars a GB but at 4 per GB its ridiculous for my dad to day PC.

If i save 5 mins a day with the better SSD or better single thread or whatever that equals to be a lot of time over a year or 2. That is 60 hours of my time and at 10 or 20 or 30 or 40 dollars per hour upgrading can easily be worth the time saved.
 
Joined
Oct 7, 2013
Messages
344 (0.09/day)
You talk about different use scenarious. Of course encryption will, as you say, hammer the drives. You are an edge case, RejZoR is closer to the avarage. You are both right, which should be obvious.
again i mentioned basic shit like downloading a game off Steam is bottlenecked by my Extreme Pro SSD...thats one of the best SATA drives on the market and it is bottlenecked. My 950 Pro is not. Through in AV/AM and a rip in the background and a downloading game? Extreme Pro and PC hangs. 950Pro? Nope...rock solid. Would it be faster with an XPoint drive? Yest it would be.

Oh also all my examples above about the Extrem Pro was also 25% OPed. The drive was 360GB. My 950 PRO has 0% OP. and still run laps and runs smoother and more consistent.

Thats all without FDE....with FDE....holy crap is it massive difference. Also even with FDE being CPU limited a faster SSD still gives better performance. I posted benchmarks in the past on hardforum about this too. I was surprised on how much more speed i got from the 950 PRO even with CPU maxed.

Added one of my initial tests below.

https://www.dropbox.com/s/kjglibxq8gzqacs/Extreme Pro vs 950 Pro Encrypted.png?dl=0
 
Last edited:
Joined
Oct 2, 2004
Messages
13,791 (1.94/day)
Oh my god, stop jumping all over the f**king place with "facts". First you talk about cache and next moment you talk how downloading a game stuffed your SSD which is used as primary drive. These things DON'T happen to SSD caches. Ever. And I'm talking about SSD caches. The entire frigging time. Coz, you know, this thread is about that...
 
Joined
Oct 7, 2013
Messages
344 (0.09/day)
Oh my god, stop jumping all over the f**king place with "facts". First you talk about cache and next moment you talk how downloading a game stuffed your SSD which is used as primary drive. These things DON'T happen to SSD caches. Ever. And I'm talking about SSD caches. The entire frigging time. Coz, you know, this thread is about that...
I was talking about entire drives the entire time. I never jumped anywhere....idiot. You claimed there is no difference from SSD cache to cheap SSD to NVMe. I clearly showed you don't know squat and showed the faster speed can easily be justified with time saved via IBMs study.

The jump from sole HDD to HDD with SSD cache is gigantic, almost identical to the level of replacing HDD with SSD entirely. The jump from crappy SSD to top end SSD is far less significant and certainly not justifiable with the price point. Even more less when that person still has HDD. Which can only mean they don't have the money. Otherwise they'd be on SSD only.

this is patently false. Proven false by IBMs study and objective data.


If you use the PC all day long and even just make 10 dollars an hour you can justify a highend SSD with your time saves assuming you actually are opening windows, programs, files, and installing..

If your APM are very low than cheap SSD is still justifiable with making only 10 dollars a day.

If you have a real job and income 960 PRO is easily justifiable with time saved.

Your biggest time saver and money maker is turning off transitions. lol
 
Last edited:
Joined
Apr 12, 2013
Messages
6,729 (1.68/day)
The only time this makes sense over a software caching solution, like Primo, is when you need tons of RAM & extremely low latency for your work. Seeing as how there aren't very many programs that can make use of micro seconds worth of latency, I'd say Intel is selling snake oil to the uninformed masses. Having used Primo, it's fairly obvious that it is the best multi level software caching solution out there, I doubt a pure hardware based implementation, which doesn't beat a RAM based cache, is gonna be too much better for the masses.
 
Top