What Performance Testing tool would you recommend

[-]

danison1337@reddit

always test on the exact software the business is using... dont go the synthetic route if you wanna have real world insights

Reply

[-]

VarmintLP@reddit (OP)

Well it's mostly to test how long the battery lasts. For the rest office and browser are the tools used on those devices ^^'

Reply

[-]

if you want to test battery life just open up youtube 24h countdown video running with full screen and full screen brightness. Then after its shut down because of the battery died plug it back in charger, open up windows and open up cmd and type powercfg /batteryreport and it should have information how the battery died and when

Reply

[-]

canadian_sysadmin@reddit

There’s usually lots and lots of public benchmarks and reviews for any given system and CPU. You shouldn’t really need to benchmark unless there’s some specific app where every percentage matters, in which case those are going to be dedicated compute machines not Surface Pros.

Reply

[-]

VarmintLP@reddit (OP)

Not the benchmark I need but a sort of CPU / GPU stress test.

Reply

[-]

FarJeweler9798@reddit

Really depends on the workflow that you want to test. Pure GPU performance, CPU Performance, single core performance, multi performance Or specific program performance like SolidWorks etc?

Reply

[-]

VarmintLP@reddit (OP)

Just something to either imitate general working. It's not exactly accurate to open Firefox and let it idle for 6 hours.

Reply

[-]

canadian_sysadmin@reddit

There’s usually lots and lots of public benchmarks and reviews for any given system and CPU. You shouldn’t really need to benchmark unless there’s some specific app where every percentage matters, in which case those are going to be dedicated compute machines not Surface Pros.

Reply

[-]

VarmintLP@reddit (OP)

Not really for a benchmark but to run some test to make the CPU work instead of just open stuff and let it idle.

Reply

[-]

paradox183@reddit

Yeah, I'm curious why a garden variety IT shop would feel running stress tests on brand new machines is worth their time in 2024.

Reply

[-]

danison1337@reddit

because if you really care the difference is big... but only if you really care obv.

Reply

[-]

dcsln@reddit

It's useful to establish a performance baseline, especially if you've standardized on hardware. Getting RAM, CPU, I/O speeds for a new device won't take a lot of time, and it gives you a point of comparison when an individual workstation "is slow". I've done this when there are dozens of matching servers, or hundreds of matching workstations. For a smaller shop, it may not be worth the effort, but that's subjective.

Reply

[-]

ConfectionCommon3518@reddit

Just stress test each unit so you can see the build quality of the unit as most stuff dies either quickly or will last its warranty length and then die the day after. Better to give a device a day's test in the torture chamber and then hand it out than have to deal with a dozen people moaning about their newly issued stuff failing and making you look stupid

Reply

[-]

VarmintLP@reddit (OP)

That's the plan with this stuff. I also need to run some tests to see if the hardware (speaker, ports, camera, etc.) works.

Reply

[-]

ConfectionCommon3518@reddit

It's going to be harder with things like usb as wear and tear, camera and speakers may need a special bit to ensure they work but generally I'd be looking at the ability to withstand x thousand throwings into the passenger seat. You can also hit problems with temps in that try and test the same device outdoors in Antarctica and Texas and see how things compare.

Reply

[-]

VarmintLP@reddit (OP)

A throwtest? I think these already got 10k throwtests behind them 😂

Reply

[-]

polypolyman@reddit

Benchmarks don't mean anything. If you're concerned about how a LOB app will perform, run tests using that app, and make sure the results are up to your standards. There's almost no reason to do burn-in testing, but a combo of prime95 and furmark should do the trick.

Reply

[-]

VarmintLP@reddit (OP)

Interesting. The surface Pros run a webtool in the browser and office. I mainly want to make sure they don't die after just a few hours compared to the 13,5hrs they would have when ne

Reply

[-]

devloz1996@reddit

My country uses PassMark as a baseline for requirements in public offers, so we also run it and keep results in asset database. I don't place my faith in benchmark results anymore though, as raw power is not the most important factor anymore - CPU instructions, TPM, RAM, disk speeds.

Reply

[-]

dcsln@reddit

HDSpeed is pretty good for simple storage read tests. IIRC, some of the write tests are destructive, so I'd use other tools for read/write, like [sqliosim](https://learn.microsoft.com/en-us/troubleshoot/sql/tools/sqliosim-utility-simulate-activity-disk-subsystem).

Reply

[-]

No_Nature_3133@reddit

BurnInTest is what I use. 3dmark is probably what you’re remembering

Reply

[-]

ORA2J@reddit

Aida64.

Reply

[-]

sysacc@reddit

I usually go with passmark to get numbers. Its not a perfect test, but it gives a good idea of the difference and you can compare other components.

Reply

[-]

oradba@reddit

Until you hone in on what you care about, Geekbench is pretty good (and free). Phoronix uses it.

Reply

What Performance Testing tool would you recommend

Reply to Post

24 Comments

danison1337@reddit

VarmintLP@reddit (OP)

FarJeweler9798@reddit

canadian_sysadmin@reddit

VarmintLP@reddit (OP)

FarJeweler9798@reddit

VarmintLP@reddit (OP)

canadian_sysadmin@reddit

VarmintLP@reddit (OP)

paradox183@reddit

danison1337@reddit

dcsln@reddit

ConfectionCommon3518@reddit

VarmintLP@reddit (OP)

ConfectionCommon3518@reddit

VarmintLP@reddit (OP)

polypolyman@reddit

VarmintLP@reddit (OP)

devloz1996@reddit

dcsln@reddit

No_Nature_3133@reddit

ORA2J@reddit

sysacc@reddit

oradba@reddit