Thunderbolt 3 M.2 NVMe SSD Enclosures

[SpecFoto]

Is anyone running 3 x 40gbps NVME enclosure on their Mac (pref M4 Pro?)

When I connect the 3rd drive it is dead slow/crashes.. (it can be any drive)

This is a known problem for Apple Silicon Macs. I have a MS M2 Max with 4 TB ports, I could only power 3 enclosures at 40Gb/s directly. (but if I intentionally unplugged 1 and later replugged in, I had to restart to see all 3 again). Seems the same for your Mac mini Pro, only 2 work at a time. It's an internal power limitation issue that all the TB ports can't provide the needed 15W each, at the same time, that external TB devices require. And that Apple doesn't like to talk about. See the link below.

https://forums.macrumors.com/threads/m4-mac-mini-external-thunderbolt-drives-limitation.2445812/

You can power additional enclosures thru a TB hub, in my case an OWC TB4 hub, but the speed is cut to around 1.2-1.4-Gb/s per port as they are sharing the TB bandwidth from 1 port on the Mac. (This speed may be increased with a TB5 hub?) I pair up 2 WD580X 4TB in separate Maiwo K1717 encosures as a RAID 0 8TB array, hooked up to the OWC hub, the speeds get back to near 2.5 Gb/s R&W, but if hooked directly as RAID 0 to the MS TB ports, speeds are 5.5R and 4.6Gb/s W The 3rd port on the hub is driving my 2nd Apple monitor.

 

[adonis3k]

Thanks mate, looks like I will be just using 2 NVME drives and my ssd which are all working fine at the moment. I was going to get a TB5 enclosure but 6x the price I paid for my 40gpbs enclosures so not worth it for me.

 

[drrich2]

It's an internal power limitation issue that all the TB ports can't provide the needed 15W each at the same time that external TB devices require.

Are there reputable brandname external single SSD enclosures with Thunderbolt 3 (or 5) that have their own power supply rather than relying on your Mac's (or a hub/dock's) Thunderbolt bus for power?

That would solve the problem, yes?

 

[adonis3k]

Maybe something like this https://www.acasis.com/products/acasis-40gbps-thunderbolt-3-m-2-nvme-2-bay-raid-ssd-enclosure or

 

[SpecFoto]

Are there reputable brandname external single SSD enclosures with Thunderbolt 3 (or 5) that have their own power supply rather than relying on your Mac's (or a hub/dock's) Thunderbolt bus for power?

That would solve the problem, yes?

Maybe something like this https://www.acasis.com/products/acasis-40gbps-thunderbolt-3-m-2-nvme-2-bay-raid-ssd-enclosure

I bought that Acasis 405 TB4 Dual NVME encolsure 3 months ago and have 2 WD Black 8TB in it as a 16TB RAID 0 array. I really like the enclosure but there are some items to be aware of:

Yes, it's self powered via a standard USB-C charger, which is great compared to most external power bricks of many others, but the Mac does not know it is self powered. It seems the power is supplied for the fan and other ports, like the monitor and USB-C 10Gb/s. So the Mac treats it as a non-powered unit and you run into the same 15W limitation and external drive limit.

There is a single cable from the Macs TB port to it, which means the 2 NVME slots are sharing the TB bandwidth and run at half speed each (Acasis claims 1.5Gb/s) with a RAID 1 backup array. If you run RAID 0 you gain back some of the speed, but not quite all of it. My Black Magic RAID 0 R/W speeds are 2.8Gb/s & 2.5Gb/s, which is in line with other dual drive enclosures in a RAID 0 array.

The fan is VERY quiet and keeps the dual drives 4 to 7 C cooler than my other single drive enclosures. Unfortunately shutting down the Mac ejects the drive and turn off the NVME's, but not the fan. You have to manually push the tiny fan On/Off switch when shutting down or the fan will run continuous. And since I don't hear the fan as the enclosure is behind my MS and Display, that happens a lot.

Still if I had a TB5 Mac, and Acasis will make a version for I, I would get it. My oldest Acasis enclosures has been trouble free for 3+ years.

 

[PaulD-UK]

@drrich2 ”Are there reputable brandname external single SSD enclosures with Thunderbolt 3… that have their own power supply…”

I have 3 Western Digital D50 TB3 docks which have an internal NVMe slot which they sell empty or fitted with an SN750 SSD.

They have two TB3 ports so are daisy chainable, plus all the usual USB ports etc.

They function absolutely reliably with my M4Pro mini, at full TB3 2800MB/s r/w speeds, although write speeds drop to ~1700MB/s if I have a 5K display plugged into the TB3 output port of the dock.

https://shop.sandisk.com/en-gb/products/ssd/external-ssd/wd-black-d50-game-dock-nvme-thunderbolt-3-ssd

Or with no SSD. However you need to provide a heatsink to fit your own choice of SSD.

0 TB WD_BLACK D50 Game Dock | Sandisk

Transform your laptop into the ultimate gaming machine with the ultra-customizable WD Black D50 Gaming Dock. Play all your favorites with new connectivity. Shop now.

shop.sandisk.com

 

[victory_over_all]

I've asked this question over on the Thunderbolt subreddit, but it looks like there's a lot of knowledge and experience about NVMe enclosures over here. Thought it might be the best place to ask!

I'm looking for the best enclosure for a 2017 iMac boot drive (Intel-based, with TB3 ports). I've settled on the Crucial T500 1TB for the drive itself, as it's DRAM and apparently has good thermals.

I'm stuck on the enclosure though. I'm not sure if I should go for the Acasis TBU405 ProM1 or Qwiizlab ES40UR.

Does the Qwiizlab ES40UR (aka Hagibis/Colorii MC40) have a true 40Gb/s "Thunderbolt 3 mode" with things like TRIM support, or does it simply fall back to a 10Gb/s USB mode when connected to a TB3 port?

Even if the Qwiizlab does have a TB3 mode, would the Acasis TBU405 ProM1 (JHL7440) be a more reliable option for a long term boot drive?

(P.S. It's for my wife's work machine, and she has an aversion to finicky tech, hence why I'm intent of finding a reliable solution that'll just work )

 

[PaulD-UK]

@victory_over_all The Qwiizlab ES40UR/Hagibis/Colorii MC40 will work with an Intel Mac, but the write speed will be half what a TB3 enclosure will achieve - about 1400MB/s in my experience.

The best results I've found are with a proper certified TB3 enclosure, ~2800MB/s R/W.
I use a Western Digital D50 gaming dock enclosure, which comes with WD SSDs installed, or you can buy it empty to fit your own SSD (a heatsink is needed).

It is on sale at reduced prices in Europe at the moment.
This works very well with my Intel iMacs, as well as with Apple silicon Macs:

https://shop.sandisk.com/fr-fr/products/ssd/external-ssd/wd-black-d50-game-dock-nvme-thunderbolt-3-ssd?sku=WDBA3U0010BBK-EESN

0 To WD_BLACK D50 Game Dock | Sandisk

Transform your laptop into the ultimate gaming machine with the ultra-customizable WD Black D50 Gaming Dock. Play all your favorites with new connectivity. Shop now.

shop.sandisk.com

 

[victory_over_all]

@PaulD-UK Ooof that WD dock looks niiiice, and those price reductions are huge! My gamer brain wants it... Cost wise it's a bit over budget though, even with the sale.

That was my worry with the Qwiizlab and other ASM2464PD-based enclosures.

In that case, I'll probably go with the Acasis. Not sure how effective the fan is, but I'm guessing the fan-cooled version is better.

Thanks for the help 👍

 

[SpecFoto]

@victory_over_all
Have been using the Acasis TBU405 ProM1 for a few months now and it is the best Acasis enclosure yet of the 3 Acasis I've had over the last 3 years (Also have experience with Maiwo K1717), The fan is ultra quiet and turns off when the enclosure is powered down. Generally it runs 3 to 5C cooler than my other non-fan enclosures. Before I bought my M2 Studio Max about 18 months ago I had a 2019 iMac and used the original Acasis drive as the startup.

Far as the 1TB NVMe, I have a Crucial 4TB but am very disappointed with it, as the cache size for the drive is way less than half what my WD 850X Blacks are. Just this morning I transferred a 988 GB file (Almost a TB) from the 8TB Black in the 405 ProM1 SU drive to the 2 Maiwo enclosures each with a 4TB WD Black, set up as a 8TB RAID. It took less than 5 minutes and the transfer speed never slowed down. With the Crucial after about 280GB of transfer data, its speed drops dramatically from 2,400Mb/s to 100Mb/s!! It started out saying it was a 6 minute process, 3 minutes in it dropped the speed and said it was a 15 minute process. You should check the T500 specs to see what size its cache is!

 

[victory_over_all]

@SpecFoto That's great to know, thanks. I'm especially reassured to know you've used it with an Intel iMac as a startup.

In terms of the drive, the Crucial T500 1TB apparently has an SLC cache of 720GB, and the WD SN850X is approx. 300GB. The T500 seems to review pretty well, and is great on paper.

I think I have my winning combo then! Acasis TBU405 ProM1 + Crucial T500 1TB.

Thanks for the help!

 

[SpecFoto]

@SpecFoto That's great to know, thanks. I'm especially reassured to know you've used it with an Intel iMac as a startup.

In terms of the drive, the Crucial T500 1TB apparently has an SLC cache of 720GB, and the WD SN850X is approx. 300GB. The T500 seems to review pretty well, and is great on paper.

I think I have my winning combo then! Acasis TBU405 ProM1 + Crucial T500 1TB.

Thanks for the help!

I certainly have a different take on the video you linked. First off the specs in the video posted at the 2:00 minute mark says the SLC cache is "up to 20% of the drives capacity" not 720 GB you state, but possibly 200GB. The video tester received numerous complaints after posting the review about the T500 substained R/W performance. They investigated and in discussing it with Crucial they were told "Crucial decided to push for as much SLC caching and as much "real world" performance as possible, at the cost of worse sustained performance when you do run out of SLC space. But this really only happens if you don't give the drive a chance to clean up (SLC -> TLC) at all, so you basically have to write 1/3rd of the capacity without any interruptions for you to see that drop." Uhh OK, so during a sustained write, we are supposed to pause it every so often to let the SLC catch catch up? How long and how many times??

I realize maybe you or your wife won't encounter this scenario, but I do as I work with very large files. The transfer I did this morning with my WD Blacks on 988GB of data took just less than 5 minutes. The same transfer with the same TB4 port and the Crucial drive in a Acasis USB4 enclosure, just to see how much slower it is, took 56 minutes......when it initially said 6 minutes..... The average speed in the end was 94MB/s Written because the SLC cache was constantly full. Oh and the temp during the transfer in the non-fan Acasis box were 64C, pretty hot vs. the 405Pro M1 at 40C. Crucial is on my No Buy list for these reasons.

 

[victory_over_all]

@SpecFoto I got the spec on the cache size from here:
https://www.techpowerup.com/ssd-specs/crucial-t500-1-tb.d1771

But on searching for the drive’s spec on another site, I found this one that says it’s approx 300GB, like the WD:
https://ssd-tester.com/crucial_t500_1tb.html

I’m pretty new to the world of NVMe drives (only started looking into it as this boot drive solution) so I’m not sure which site would be correct. Maybe the 720GB figure is for a larger capacity, and not the 1TB like claimed on that first link above?

But damn that drop in performance during big transfers is not good! My wife doesn’t often deal in multi-GB file transfers, but it’s not great knowing that the issue is there.

I didn’t quite grasp how much of an issue that could be, so thanks for pointing it out.

The SN850X is only €7 more than the T500, so I’ll look into that drive instead. Good to know that the thermals on the WD are good too, particularly in the Acasis enclosure. I see a lot of complaints about NVMe enclosures getting hot, with the enclosure being blamed, but I thought that the drive probably plays a big role here, hence my concern about thermals.

 

[SpecFoto]

@victory_over_all

The tech power specs are for all versions but of course they post the largest, and fastest, version for the results.

Just to make it clear my Crucial NVMe is not the T500, but the P3 Plus Gen 4 drive bought in June of 2024. From what I read it's a step below the T500 in performance. Normally it's just a bit slower than the WD 850X Black, my Mac told me 5 minutes for the WD 1 TB file transfer and then 6 minutes total for the P3 Plus, that is until this artificial limit of filling the buffer hits then it's a slow as a HDD at approximately 1 hour/TB as my test today shows. I did not find out about the smaller buffer and the speed limitation until after the 30 day period for a return was up. The 4TB and 8TB versions of the WD Black 850x I use seem to have much larger buffers, more than 1TB, and I have not had a slowdown issue. The P3 Plus has been regulated to a 3rd backup drive where I can let it run overnight if needed.

 

[Elviss116]

new update 54.17:

 

[EugW]

I got an LG UltraFine 6K display with integrated Thunderbolt 5, although I'm running it with my Thunderbolt 4 M4 Mac mini.
I also have a Plugable Thunderbolt 4 / USB 4 hub, and a Qwiizlab fanless ASM2464PD enclosure with Samsung 990 Pro SSD.

With the drive connected directly to the Mac, it gets detected as USB 4, runs higher power, at fastest speed.

With the drive connected through the Plugable hub, it gets detected as TB, runs lower power, at a bit slower speed.

When the drive is connected through the LG 6K monitor, it gets detected as USB 4, runs higher power. Writes half speed.

 

[Pressure]

I got an LG UltraFine 6K display with integrated Thunderbolt 5, although I'm running it with my Thunderbolt 4 M4 Mac mini.
I also have a Plugable Thunderbolt 4 / USB 4 hub, and a Qwiizlab fanless ASM2464PD enclosure with Samsung 990 Pro SSD.

With the drive connected directly to the Mac, it gets detected as USB 4, runs higher power, at fastest speed.

View attachment 2574693

View attachment 2574692

With the drive connected through the Plugable hub, it gets detected as TB, runs lower power, at a bit slower speed.

View attachment 2574695

View attachment 2574694

When the drive is connected through the LG 6K monitor, it gets detected as USB 4, runs higher power. Writes half speed.

View attachment 2574697

View attachment 2574696

You can resize the image sizes 😂

 

[PaulD-UK]

@EugW ”When the drive is connected through the LG 6K monitor, it gets detected as USB 4, runs higher power. Writes half speed.”

That is normal, expected, behaviour - for a TB4 connection.
Because there is only ~40Gbps (nominal) bandwidth, so the video DSC data stream bandwidth is not available for the USB4 data storage Write stream.

With a TB5 Mac, there should be no slow down.
It’s quite possible the SSD would get about ~4-5Gbps better performance - it would if connected directly to a TB5 Mac, so might also when connected to the TB5 display?

 

[joevt]

I got an LG UltraFine 6K display with integrated Thunderbolt 5, although I'm running it with my Thunderbolt 4 M4 Mac mini.
I also have a Plugable Thunderbolt 4 / USB 4 hub, and a Qwiizlab fanless ASM2464PD enclosure with Samsung 990 Pro SSD.

With the drive connected directly to the Mac, it gets detected as USB 4, runs higher power, at fastest speed.

With the drive connected through the Plugable hub, it gets detected as TB, runs lower power, at a bit slower speed.

When the drive is connected through the LG 6K monitor, it gets detected as USB 4, runs higher power. Writes half speed.

6K using DSC@12bpp requires ≈16 Gbps so it makes sense that connecting the SSD enclosure to the LG 6K would reduce the write speed.

But it seems that the write speed is more reduced than I would expect.
40 Gbps - 16 Gbps = 24 Gbps.
1444.4 MB/s = 11555.2 Gbps (much less than 24 Gbps)

What's the pixel clock of the display? Maybe it's greater than what is required by CVT-RB timing calculation.
What's the DSC compression? Maybe it's more than Apple's usual 12bpp.

What does AmorphousDiskMark.app say? Maybe Disk Speed test has more overhead?

 

[EugW]

@EugW ”When the drive is connected through the LG 6K monitor, it gets detected as USB 4, runs higher power. Writes half speed.”

That is normal, expected, behaviour - for a TB4 connection.
Because there is only ~40Gbps (nominal) bandwidth, so the video DSC data stream bandwidth is not available for the USB4 data storage Write stream.

With a TB5 Mac, there should be no slow down.
It’s quite possible the SSD would get about ~4-5Gbps better performance - it would if connected directly to a TB5 Mac, so might also when connected to the TB5 display?

Yes I figured, since it’s TB 4.

I was actually pleasantly surprised it worked so well through the monitor, and was also curious about the drive mode detection.

Direct to Mac - Drive seen as USB 4
Thru TB 4 hub - Drive seen as TB 3
Thru monitor - Drive seen as USB 4

This matters to me because I actually prefer the slightly slower TB mode as the drive will idle at lower power. However, I will also need to check the active wattage, given the lower write speed, but idle is more important here since drive is usually idle, housing my Photos Library.

6K using DSC@12bpp requires ≈16 Gbps so it makes sense that connecting the SSD enclosure to the LG 6K would reduce the write speed.

But it seems that the write speed is more reduced than I would expect.
40 Gbps - 16 Gbps = 24 Gbps.
1444.4 MB/s = 11555.2 Gbps (much less than 24 Gbps)

What's the pixel clock of the display? Maybe it's greater than what is required by CVT-RB timing calculation.
What's the DSC compression? Maybe it's more than Apple's usual 12bpp.

What does AmorphousDiskMark.app say? Maybe Disk Speed test has more overhead?

IIRC, ADM is usually roughly around 10% faster for sequential.

This is a Samsung 990 Pro so the bottleneck in sequential speeds is the enclosure. It’s been a long time since I benchmarked it with ADM but off the top of my head, in directly connected USB 4 mode, I think it maxed out around 3500ish MB/s.

3500x8 = 28 Gbps

16 Gbps DSC + 12 Gbps SSD = 28 Gbps

Isn’t that in the right ballpark? I thought the max real data rate is about 32 Gbps but that would include additional overhead for the SSD.

BTW, is it correct to assume the data bandwidth is shared with the monitor across the four 8 Gbps lanes? I’m assuming the drive doesn’t get its own two lanes.

How do I determine the monitor bandwidth stats in DSC? Also, wouldn’t DSC be turned off on a Thunderbolt 5 Mac? DSC is not needed there.

PS. Monitor was calibrated for Display P3 (or strictly speaking, P3 wide colour, 6500K white point, gamma 2.2 since there is no Display P3 preset in the LG Calibration Studio software), and was running in SDR mode, not HDR.

EDIT:

Found an old ADM and Black Magic bench of this drive from last year, when the drive was empty.

3450 MB/s / 27.6 Gbps vs 3204 MB/s so a difference of 8%

 

[joevt]

This is a Samsung 990 Pro so the bottleneck in sequential speeds is the enclosure. It’s been a long time since I benchmarked it with ADM but off the top of my head, in directly connected USB 4 mode, I think it maxed out around 3500ish MB/s.

3500x8 = 28 Gbps

16 Gbps DSC + 12 Gbps SSD = 28 Gbps

Isn’t that in the right ballpark? I thought the max real data rate is about 32 Gbps but that would include additional overhead for the SSD.

Yeah but DisplayPort is usually counted separate from data.
For example, an Apple XDR display using dual tile mode instead of DSC can use 38 Gbps of Thunderbolt 3.

BTW, is it correct to assume the data bandwidth is shared with the monitor across the four 8 Gbps lanes? I’m assuming the drive doesn’t get its own two lanes.

Thunderbolt3/4 and USB4 is 2 lanes of 20 Gbps. 40 Gbps total each direction. DisplayPort, PCIe, and USB data are encapsulated inside Thunderbolt packets - like how Ethernet can transmit HTTP and FTP and SMB. The Thunderbolt controllers at each end of a Thunderbolt connection (host and peripheral) have PCIe, DisplayPort, and USB Up and Down (or In and Out) adapters to do the conversion to and from Thunderbolt. The DisplayPort, PCIe, or USB, are said to be "tunnelled".

How do I determine the monitor bandwidth stats in DSC?

We can interpret the Color modes in the ioreg that are supported by the current display mode.
I don't know if the BetterDisplay developers have found a method or API to find the current color mode for the current display mode.

Also, wouldn’t DSC be turned off on a Thunderbolt 5 Mac? DSC is not needed there.

If both the display and the Mac supports DisplayPort 2.1 UHBR 13.5, then it could support 6144x3456 60Hz 10 bpc RGB without DSC but that might be a waste of bandwidth since DSC is supposed to be visually lossless.

Connect the display to an Intel Mac and use AllRez to get DisplayPort capabilities. I have not searched for a method or API to get the same info on Apple Silicon.

 

[EugW]

If both the display and the Mac supports DisplayPort 2.1 UHBR 13.5, then it could support 6144x3456 60Hz 10 bpc RGB without DSC but that might be a waste of bandwidth since DSC is supposed to be visually lossless.

The LG 6K is DP 2.1 UHBR 13.5, according to LG's own specs.

Connect the display to an Intel Mac and use AllRez to get DisplayPort capabilities. I have not searched for a method or API to get the same info on Apple Silicon.

I downloaded your Allrez app, and ran it on my 2017 m3 MacBook with the LG 6K. Is this old anemic Mac enough to provide sufficient info?

It claims to be able to output 6720x3780 or HiDPI 3360x1890 or HiDPI 3008x1692, but that's not what the display is receiving, since if I turn on the monitor's "Just Scan" setting (for 1:1 pixel display), only the middle part of the screen is being used. I believe it is scaled down to 3840x2160 for output. Interestingly, there is no option for 3072x1728.



However, the screen grabs of 6720x3780 are indeed 6720x3780.



Anyhow, I have attached the Allrez output (without the serial number).

 

[joevt]

The LG 6K is DP 2.1 UHBR 13.5, according to LG's own specs.

I downloaded your Allrez app, and ran it on my 2017 m3 MacBook with the LG 6K. Is this old anemic Mac enough to provide sufficient info?

m3 in this case means 1.2GHz dual-core Intel Core m3.
We just want to look at the DisplayPort info, so this should be sufficient.

The DisplayPort info reports the following:

                        Receiver Capability
                            00060h DSC_SUPPORT: DSC_DECOMPRESSION_IS_SUPPORTED
                            00061h DSC_REV: 1.2
                            00062h DSC_RC_BUF_BLK_SIZE: 64kB
                            00063h DSC_RC_BUF_SIZE: 127 * DSC_RC_BUF_BLK_SIZE
                            00064h DSC_SLICE_CAP_1 & 2: 1, 2, 4, 6, 8 max slices per DisplayPort DSC sink
                            00065h DSC_LINE_BUF_BIT_DEPTH: 16 bits
                            00066h DSC_BLK_PREDICTION_SUPPORT: DSC_BLK_PREDICTION_IS_SUPPORTED
                            00067h DSC_MAX_BITS_PER_PIXEL: 0 bpp
                            00069h DSC_DEC_COLOR_FORMAT_CAP: RGB, YCbCr 4:4:4, YCbCr 4:2:2 Simple, YCbCr 4:2:2 Native, YCbCr 4:2:0 Native
                            0006ah DSC_DEC_COLOR_DEPTH_CAP: 8, 10, 12 bpc
                            0006bh DSC_PEAK_THROUGHPUT: MODE_0 = 340 Mp/s, MODE_1 = 340 Mp/s
                            0006ch DSC_MAX_SLICE_WIDTH: 2560 pixels
                            0006fh DSC_BITS_PER_PIXEL_INC: 1 bpp
                            00090h FEC_CAPABILITY: FEC_CAPABLE, FEC_UNCORR_BLK_ERROR_COUNT_CAP, FEC_CORR_BLK_ERROR_COUNT_CAP, FEC_BIT_ERROR_COUNT_CAP, PARITY_BLOCK_ERROR_COUNT_CAP, PARITY_ERROR_COUNT_CAP, FEC_ERROR_REPORTING_POLICY_SUPPORTED, ?0x40

                        Extended Receiver Capability
                            02215h 128B132B_SUPPORTED_LINK_RATES: UHBR 10, UHBR 13.5

Your 2017 MacBook is limited to HBR2 link rate from Intel HD Graphics 615, so the max pixel clock is < 720 MHz for 8bpc RGB or < 576 MHz for 10 bpc RGB.
In order for HBR2 to get 6K60 without DSC, the GPU would need to support 8 bpc 4:2:0 and 1345.11 MHz.
I think IOFBTimingRange shows a limit of 650 MHz.

It claims to be able to output 6720x3780 or HiDPI 3360x1890 or HiDPI 3008x1692, but that's not what the display is receiving, since if I turn on the monitor's "Just Scan" setting (for 1:1 pixel display), only the middle part of the screen is being used. I believe it is scaled down to 3840x2160 for output. Interestingly, there is no option for 3072x1728.

However, the screen grabs of 6720x3780 are indeed 6720x3780.

The list of display modes shows a 4K 60Hz mode using HDMI timing of 594 MHz. Most modes are scaled to 4K 60Hz CVT-RB timing (533 MHz).
All the modes > 4K are scaled down to the 4K 533 MHz mode. You need a utility like SwitchResX to see that a mode is scaled down or not.