Memory Mark Asus Zephyrus G14 (2025)

[Turbo12R]

The Memory Mark test seems unusually low considering the memory specs on this laptop. Is something being measured incorrectly?

[David (PassMark)]

You need to post more details before anyone can attempt to provide a sensible answer.

What software are you using?
What was the memory mark score?
Is the result reproducible over several runs?
How much RAM is in the G14 and the layout (if you know it, e.g. 2 sticks in dual channel).
What clock speed is the RAM running at?
Was this tested on mains power or battery power?

[Turbo12R]

Investigating my system itself as I think the benchmark is working as intended, I've looked at similar systems on here though and it seems they are getting roughly the same benchmarks. Specs on the memory are 32GB LPDDR5X 8000, CPUz however reports 4x32 @ 998 MHz, which I guess would be 4000 MHz, half of what the specs say. Maybe I am thinking wrong on something here..,? but that would also explain why I am thinking the memory benchmark is slow. Link to my Passmark run: (tested on main power)

https://www.passmark.com/baselines/V...d=268960113692

[David (PassMark)]

I looked up the spec for this Asus Zephyrus G14 (2025). What I found was this claim for RAM.
"32GB LPDDR5X 7500 on board (Actual memory speeds may vary by CPU configuration.)"

I am sure that they add this "Actual memory speed" disclaimer because the RAM probably doesn't actually runs at 7500MT/s.

Most of the memory benchmarks weren't all that bad. Latency was the worst by far however. This might indicate they have setup the RAM timings for low power usage and not max performance. The purpose of LP (low power) RAM after all is to save power.

CPUz however reports 4x32 @ 998 MHz

This doesn't make sense to me. 4 x 32 would normally means 4 slots with 32GB (so 128GB). Which isn't right.
998 MHz is also way too slow for DDR5.
But beware. Modern laptops ramp their RAM clock speeds up and down when under load. The clock speed is variable.

[Turbo12R]

Ya, mine is actually the WW model at 8000, but close enough. Timings are kind of bouncing all over the place, and I am in high performance profile.

[David (PassMark)]

4 x 32

From the screen shot I can see what you are talking about now. DDR5 has two 32bit channels. Memory controller also has 2 channels. So 4 channels. This is not related to the clock speed or MT/s value however.

[Turbo12R]

I guess I was thinking for some reason the Channel # were representative of each stick of memory (4 sticks) and then that number was multiplied by the indicated DRAM frequency (998 in this case) to get the total frequency, which would be about 4000. I guess that's where I am getting confused as I would have thought the DRAM frequency should have been about 2000 * 4 channels = 8000 (rated frequency)

Windows does report it at 8000 though:

[David (PassMark)]

I'm fairly sure the value in task manager is the maximum configured speed, not the current real-time speed.

[Turbo12R]

I'm fairly sure the value in task manager is the maximum configured speed, not the current real-time speed.

I agree. Oh well, the tested speed is slower than I was hoping, but on a whole for the form factor of this laptop it runs very well. If I did not see the memory benchmark, I would never guess it was maybe a little on the slow side. Thanks for your input.

[David (PassMark)]

The 8000MT/s speed is actually very high (as at June 2025).

But by only looking at the clock speed, you are only getting half the story. The timings also play a big part. These are 4 numbers like, 38-50-50-128. They are known as the CAS latency (CL), tRCD, tRP, and tRA values. They are measured in clock cycles.

tRCD - Row Address to Column Address Delay
tRP - Row Precharge Time
tRAS - Row Active Time

So for example, 38 (CAS Latency, CL) means it will take 38 clock cycles for the RAM to respond to a read command after receiving it.
So high clock speed doesn't mean too much, unless the timing numbers are also reasonable.