Question Get-FileHash benchmark with Powershell

[igor_kavinski]

Downloaded the Win11 24H2 US-English ISO (5.4GB).

Opened Powershell. It will usually open in the current user folder. Typed CD Downloads to make that the current folder.

Typed this: Measure-Command {Get-FileHash *.iso} <<< (just make sure there are no other ISOs in there, or just create a separate folder and run the command in that folder).

RESULTS:

Core 14700K Optane 1600X: 3.887 seconds

Ryzen 7950X3D Samsung 990 Pro dual channel DDR5-6000C30: 3.889 seconds

Apple M1 base MBA: 4.016 seconds

Ryzen 7800X3D Ramdrive: 4.844 seconds

Ryzen 7700X Kingston KC3000 dual channel DDR5-6000: 5.117 seconds

Ryzen 5800X3D Sabrent NVMe dual channel DDR4-3600: 6.196 seconds

Core i3-1125G4: 7.071 seconds

Core i5-12500 Kioxia NVMe SSD dual channel DDR4-3200: 8.8 seconds

Core i7-4790K SATA SSD (probably dual channel) DDR3: 26 seconds

Core i5-10500 Samsung 870 EVO 250GB SSD single channel DDR4-2666: 27.340 seconds

Core i7-4770 Samsung 500GB SATA SSD dual channel DDR3-1600: 35.74 seconds

Core i5-3570 MX500 250GB SSD 4GB single channel DDR3-1600: 41.052 seconds

2x Xeon E5-2630 V2 Dell PERC H310 RAIDed (probably RAID 1) 10K RPM HDD (no idea how many channels of RAM): 76.483 seconds


I would love to see scores of Zen 1, 2, 3, 4, 5 and Raptor/Meteor/ADL-N/Lunar and Arrow Lake, though don't expect to see anything less than 3 seconds because I'm assuming it is quite I/O bound and SSD/RAM speed may play extensive role in the process.

 

[igor_kavinski]

Discovered the CertUtil command when I tried to run the benchmark on Windows 7 as it doesn't have the get-filehash command.

Ran these commands:

Measure-Command {CertUtil -hashfile .\Win11_24H2_English_x64.iso SHA256}
Measure-Command {CertUtil -hashfile .\Win11_24H2_English_x64.iso SHA512}
Measure-Command {CertUtil -hashfile .\Win11_24H2_English_x64.iso MD5}

 

[SK10H]

I would love to see scores of Zen 1, 2, 3, 4, 5 and Raptor/Meteor/ADL-N/Lunar and Arrow Lake, though don't expect to see anything less than 3 seconds because I'm assuming it is quite I/O bound and SSD/RAM speed may play extensive role in the process.

7800x3d ramdisk. Higher clocked ccd should beat 4 secs, or port those GPU/ASIC sha256 miners to do it.

 

[igor_kavinski]

Higher clocked ccd should beat 4 secs

Now I'm thinking Det0x will beat 3 seconds!

 

[igor_kavinski]

 

[DrMrLordX]

Is this a CPU, storage, or platform benchmark?

 

[igor_kavinski]

Is this a CPU, storage, or platform benchmark?

Read intensive for storage but once the data gets into RAM, the CPU has to crunch through it really quickly and there's a lot of read/write activity in the allocated RAM so RAM bandwidth is also important. We need a 7700X score to understand what part the cache plays here.

By the way, I think this is an ST workload because I didn't see any multicore activity.

 

[iafro89]

14700k, Intel Optane P1600X 118 GB as DevDrive




edit* redone with dir output

 

[igor_kavinski]

14700k, Intel Optane P1600X 118 GB as DevDrive

Can you please also do the CertUtil benches? What's your RAM speed/timings?

 

[igor_kavinski]

 

[igor_kavinski]

@Nothingness @Gideon

Would you guys like to contribute your scores, with your beefy M1 Max, M2, M3 etc. CPUs?

 

[poke01]

AMD 7700X
32GB DDR5 6000MT/s
1TB KC3000


5.11

 

[Sunaiac]

5800x3D, 3600CL15 RAM, from Gen 3 NVME sabrent, Win 11.

 

[Hail The Brain Slug]

7950X3D, stock
6000C30 EXPO 2x32GB, untuned
990 Pro 4TB

3.889s

 

[igor_kavinski]

Weirdly, I don't think this benchmark depends that much on storage speed. That i3-1125G4 score improved by about a second when I ran it immediately again from a USB flash drive with max read speed of 30 MB/s. I think the cached data may also have contributed to that speedy score. Will need to check again on that UFD with a cold run.

 

[Sunaiac]

The score I gave you was on second run right after a first one that was not even 0.3s slower. (DDR4 3600 vs PCIe Gen3 NVME)

 

[Hail The Brain Slug]

Weirdly, I don't think this benchmark depends that much on storage speed. That i3-1125G4 score improved by about a second when I ran it immediately again from a USB flash drive with max read speed of 30 MB/s. I think the cached data may also have contributed to that speedy score. Will need to check again on that UFD with a cold run.

You can check in task manager the standby memory consumption. From a fresh boot, there should be very little in memory. If you run the benchmark, see if the standby allocation increases by 5.4GB afterwards. That would indicate windows has loaded the file in memory and not freed it, leaving it effectively cached until there's a compelling reason to free up the capacity.

I suspect if you run it from that USB from a fresh boot, it will be really, really slow.

 

[Hail The Brain Slug]

I just tested it. The file does live in standby memory after an initial run.

Also my system has decided it only takes 3.706s now. lol.

 

[igor_kavinski]

Also my system has decided it only takes 3.706s now. lol.

Is that the first run or 2nd run after the file is in RAM?

 

[Ranulf]

win1124H2 english iso

4790k 4.0ghz, sata tlc 2tb ssd, (while watching a yt stream)

26s

 

[igor_kavinski]

win1124H2 english iso

4790k 4.0ghz, sata tlc 2tb ssd, (while watching a yt stream)

26s

You changed the parameters of the benchmark. Now everyone has to watch your particular YT stream and revise their benchmarks while doing so

But your score is certainly better than my "i7-4770 without watching YT" so I guess it's valid for the leaderboard.

 

[Ranulf]

It was minimized at least. Heh.

 

[coercitiv]

First run takes significantly longer, after which I guess the file ends up in Windows RAM cache. For the test below I even kept the task manager open to see the cache inflate by ~5GB in real time

12700K DDR4 3600, 990 Pro. IIRC the first run I did after downloading the file was around 5 seconds, which suggests it had already been cached.

 

[Hail The Brain Slug]

Is that the first run or 2nd run after the file is in RAM?

Second run from ram, which appears to also be what happens when you first download the ISO - its already in standby memory so you aren't testing the disk.

 

[igor_kavinski]

That "already cached" when downloaded part throws a wrench into the reliability of this benchmark then

Need to revisit some of my machines to see how a run after a fresh boot fares...