Microsoft is Switching from MHz to MT/s in Task Manager for Measuring RAM Speeds

[AleksandarK]

The battle is over. Microsoft is finally changing the measuring methodology in its Task Manager from Mega Hertz (MHz) to Mega Transfers per second (MT/s). This comes amid the industry push for more technical correctness in RAM measuring, where the MHz nomenclature does not technically represent the speed at which the memory is actually running. While DRAM manufacturers list both MHz and MT/s, the advertised MHz number is much higher than the effective speed at which the DRAM is running, resulting in confusion and arguments in the industry about choosing the correct labeling of DRAM. A little history lesson teaches us that when single data rate (SDR) RAM was introduced, 100 MHz memory would perform 100 MT/s. However, when double data rate (DDR) memory appeared, it would allow for two memory transfers per clock cycle.

This would introduce some confusion where the MHz speed is often mixed up with MT/s. Hence, Microsoft is trying to repair the damage and list memory speeds in MT/s. Modern DDR5 memory makers are advertising DDR5 kits with "DDR5-4800" or "DDR5-6000," without any suffix like MHz or MT/s. This is because, for example, a DDR5-6000 kit runs at 6,000 MT/s, the effective speed is only 3,000 MHz. The actual clock of the memory is only half of what is advertised. The MT/s terminology would be more accurate and describe memory better. This Task Manager update is in the Windows 11 Insider Preview Build 22635.3570 in the Beta Channel, which will trickle down to stable Windows 11 updates for everyone soon.



View at TechPowerUp Main Site | Source

 

[Athlonite]

I always love the look on the faces of people who tell me the speed of their ram when I tell them they're wrong and explain how it really is

it's double data rate not double the MHz speed

 

[Assimilator]

Why can't they just display both?

 

[Onasi]

Why can't they just display both?

I am sure it’s some nonsensical thing about clean design and minimizing number clutter. This is honestly probably the least needed change to Task Manager currently. The “new” TM is still just a graphical “modern” shell over the same program code that’s been there since Vista (if not 2000/XP) which makes it run like absolute ass and have hilarious behaviors like updating the window at 60hz regardless of monitor refresh. Great GPU accelerated modern GUI there MS, very poggers of you. It’s been this way for more than a year now. But apparently fixing a terminology gaff takes precedence over actually insuring the thing works properly.

 

[Tomorrow]

MHz and especially MT/s (that no other memory standard uses) are meaningless numbers. Just display the raw bandwidth in Mbps and case closed.
Also:

MT/s is a horrible unit. Mbps is the SUPERIOR unit for refering to effective memory speeds

 

[ExcuseMeWtf]

I approve of this move.

 

[Wirko]

MT/s (that no other memory standard uses)

Except LPDDR, GDDR, and the interface between NAND chips and SSD controller (which is also DDR). And add PCIe on top of that.

But in current interfaces, at least MT/s is the same as Mbit/s (or Mbps, wherein "p" stands for profane). One bit per transfer. Some new interfaces will have 1.5 or 2 bits per transfer. That's where MT/s becomes impractical, so we'll have to use "Mbit/s per pin".

 

[Tomorrow]

Except LPDDR, GDDR, and the interface between NAND chips and SSD controller (which is also DDR). And add PCIe on top of that.

But in current interfaces, at least MT/s is the same as Mbit/s (or Mbps, wherein "p" stands for profane). One bit per transfer. Some new interfaces will have 1.5 or 2 bits per transfer. That's where MT/s becomes impractical, so we'll have to use "Mbit/s per pin".

Manufacturers of said technologies themselves dont use MT/s so why should we?
Like i said: raw bandwidth is universal and easily understandable.

 

[Wirko]

updating the window at 60hz regardless of monitor refresh

How do you get it to update that fast? Mine (Win 7 and Win 11) update at 2 Hz or so.

 

[WhateverAnotherFreakingID]

Mega Transfers

are these some Imperial System witchery?

 

[Onasi]

How do you get it to update that fast? Mine (Win 7 and Win 11) update at 2 Hz or so.

You mean the process statuses themselves, how often the TM polls the system? Yeah, those are slow to update, although I believe there’s an option (at least on 11 version) to increase the frequency of polls at the coat of CPU overhead.
Wasn’t talking about those though, I referred to GUI itself in my post. Could easily be noticed on a higher refresh screen while, for example, scrolling the process extended list or services up and down. I agree that it doesn’t even feel like 60, that probably is due to how messed up the MS new UI is. But the old TM which this still is underneath the shell SHOULD work at 60. How much slower the transition process made it I can’t say, but it is a slow and laggy POS.

 

[R0H1T]

How do you get it to update that fast? Mine (Win 7 and Win 11) update at 2 Hz or so.

You mean 2 MT/s ?

 

[qlum]

I wonder if at some point the start correctly reporting storage sizes in windows.
Instead of Misreporting the Base 1024 KiB MiB GiB... as KB MB GB, either move the reporting to base 1000 or use the correct name.

 

[_JP_]

Waiting for the "CoPilot enabled core" count stat!

 

[Wirko]

I wonder if at some point the start correctly reporting storage sizes in windows.
Instead of Misreporting the Base 1024 KiB MiB GiB... as KB MB GB, either move the reporting to base 1000 or use the correct name.

Or, allgodsforbid, leave the choice to the user.

Wasn’t talking about those though, I referred to GUI itself in my post. Could easily be noticed on a higher refresh screen while, for example, scrolling the process extended list or services up and down. I agree that it doesn’t even feel like 60, that probably is due to how messed up the MS new UI is. But the old TM which this still is underneath the shell SHOULD work at 60. How much slower the transition process made it I can’t say, but it is a slow and laggy POS.

Those lists are a bit special, they are being updated in real time, and if the updating is poorly implemented, it could cause lags and stutter. That's not an indication that the UI design in general is poor (but it's poor anyway).

What about file lists in file explorer? Do they scroll smoothly regardless of which view (icons or details) you choose?

Manufacturers of said technologies themselves dont use MT/s so why should we?
Like i said: raw bandwidth is universal and easily understandable.

So I checked a few examples, expecting that I'd find a weird mix, and I found a weird mix.

DDR5 modules by G.Skill, Crucial, Corsair and Kingston: MT/s
DDR5 motherboards: Asus doesn't mention units, MSI uses MHz incorrectly, Gigabyte uses MT/s
CPUs: Intel uses MT/s, AMD too for Epycs, but no units for Ryzens
DDR5 chips: Samsung uses Mbps
GDDR on 4090 GPUs: Asus and MSI say 21 Gbps, they just copy Nvidia's data
LPDDR5 in notebooks: Dell has Mbps

So raw bandwidth in gigabits per second is not at all common to describe DDR5 memory modules. It's more common in other cases. I was wrong about GDDR and LPDDR. The distinction seems to be between chips (bit/s) and modules (transfers/s).
The old metric (megabytes per second per module) appears to be dead, although JEDEC still defines ratings such as "PC5-64000".

 

[Onasi]

What about file lists in file explorer? Do they scroll smoothly regardless of which view (icons or details) you choose?

Yeah, that works fine after… what is it, the third time MS redone the Explorer in Win11? There is a slight delay on opening it initially (just a tad, still not as immediate as the old Win7 Explorer, but far cry from the initial Win11 launch explorer or the second iteration that DID stutter like mad), but from that the scrolling seems smooth and adhering to the refresh. Regardless of the view chosen, though I use List. Haven’t checked on any really big folders with hundreds or thousand, so that may misbehave, will check later. Or if there are a lot of tabs, will need to check that too, although I imagine that would just cache out to RAM and won’t cause issues if that’s sufficient.

 

[user556]

It was never any big deal, imho. Even when talking at the low level coding/config of hardware I knew what was implied no matter the term in use. Very rarely is the actual clock frequency implied when talking MHz. It's just a generic measure of rate.

 

[boomheadshot8]

Microsoft asked IA to use this type of mesure ?

 

[Wirko]

Hey, the "Slots used" field even has a tooltip with additional useless info. But the TM and the system info app are still unable to tell if the memory is DDR4 or DDR5. "3200 MHz" may mean DDR4-3200 or DDR5-6400.

 

[Sabotaged_Enigma]

Nice! Finally!

 

[Ripcord]

waste of time, it doesn't take into account latency. just changing from one inaccurate system to another.

 

[Deorc Mearh]

I completely disagree. I want to know at what frequency the piece of hw works. If it doubled the amount of data transferred at once is my decision to inquire or not. You don't read cpu works at 40 Ghz because when their clock is 5Ghz and they have 8 cores. We know since ddr2 memories uses techniques for sending more data keeping the channel frequency at the same value.What you need to do is comparing frequencies of different memory with the same technology: ddr5 3000MHz are faster than 2800MHz. That's all. - obviously you must compare them at the same latency settings

 

[chrcoluk]

Passmark succumbed as well, I noticed when I updated memtest86 on my multiboot ISO one of the changes was MHZ to MT.