Print

[ssp3]

DieHard, you're twisting the facts 

If you

Buy this for $13: https://www.amazon.com/mSATA-44pin-Notebook-Laptop-Enclosure/dp/B01GRMUQRG?th=1

and don't try to stick cheap, chinese, large capacity, power hungry, no-name SSD into it, you'll be absolutely fine. No problems with these + brand name, low power mSATA drives so far.

[DieHard]

Thanks for clearing that up
Sorry for muddying the waters

The heat tests that FBz will be doing will also be helpfull of which way to go

[FBz]

The AbleConn mSATA sleds are tested proven and stable. Good performers. BUT I have a personal dislike of the Marvells, based on their inherent higher cost. Always checking for lower cost, and comparable performance alternatives.

Here’s some example QB 2.0 results from a lowly 1.25 GHz G4 Mac mini last week. Not quite up to DieHard’s Ableconn in a 1.42 GHz mini (but I began recent testing with the lower 1.25 GHz mini). Similar tests from a 1.42 and a 1.5 GHz mini are re-slated. Top two results below are mSATA. Bottom two are with JMicron bridge adapters + stripped naked SSD. (Clickable image.)

   
    Hey IIO, can you import and plot performance curves from such QuickBench 2.0 generated text?

*Also ran SCSI Benchmark too, along with 1.42 & 1.5 GHz minis but that USB stick with all of those results from last week "expired".
 mSATA formatted and OS’es installed via Tiger in a Quicksilver & using Disk Utility. Intend to retest all again using Drive Setup 2.1.
 (Just as there is an overall performance difference between using Drive Setup 1.9.2 and Drive Setup 2.1.)
 All drives here (mSATA, SSD, etc.) are typically two-partitioned, dual boot (OS 9.2 & 10.4.6)

 Often, so many variables to consider.

 And, guess I’ll need to buy a newer MCA004 V1.3 to test - because the AS331 V1.5’s are no more. 


With nearly 12,000 views of this thread so far and especially considering OS 9.2 / V.9 for the G4 Mac mini… scores have checked out all of the options and the intent here was always to provide good solid working info with the various configs, formatting, etc. And of course, the proverbial “best bang for the buck”.


Thanks ssp3 for the M.2 results. I was expecting higher numbers. DieHard’s inexpensive Asenno mSATA in a 1.42 GHz Mac mini, riding his AbleConn / Marvell-based sled… coughed up an 89.226 MB/sec per QB 2.0 sequential read result (@1MByte transfer size) last week.


“I look at these two boards and am wondering how do they work at all.” -ssp3

Same goes for the “Bribge” EVB-002-3 ehh? It shouldn’t - but it does work so very well. 

And thanks to ssp3 for checking out the JMicron images and his insights.
Perhaps like some things around here, we’re occasionally surprised by things that shouldn’t work (but do) and by things that should work (but do not). Of course we all appreciate more, the former… rather than the latter.


There are still six remaining G4 Minis here with the V1.5’s - and two loose V1.5’s for testing. [I should’ve bought more!] Hopefully the “white case” adapters will be a suitable low-cost replacement. I have seen some mSATA upper heat limit specs noted as 70˚ C / 158˚ F. And while I’ve yet to ever approach that threshold even in a bone stock G4 Mini with HDD… I’m never concerned in any Mac until surpassing 138˚ F. (Except for maybe the insanely hot running first Core2Duo Minis!)

The G4 Minis ultimately run much cooler, as a rule.

Last week initial tests here with the two “white case / box” adapters (naked): one did produce some oddly rising temps on specific PCB traces. (More so than on their JM2330’s or voltage regulators.) Nothing above 109˚… however, after the first 10-15 minutes in an open case mini. I intend to check this more closely and under longer run times (top on & top off).

UND the AbleConn sled too…

Ahh, testing continues...
But now it is afternoon here and I haven't had coffee OR breakfast.
(Likely will be days before I return to this and more testing.)

доверяй, но проверяй. 

[akator]

I've got that same $13 adapter in a couple of PPC Minis.  I've got even cheaper (~$8) SATA to IDE adapters in a couple of other PPC Macs, one a Mini and another an iMac.  I haven't had any problems in the last several years that these have been used.

All of my Macs, Mini or otherwise, have reduced temps with SSDs installed, even using the less expensive adapters and regardless of what SSD is used.

None of these PPCs can even get remotely close to maximizing the speeds on these SSDs. 

[ssp3]

доверяй, но проверяй. 

Huh? You one of Z-em?

[FBz]

G4 Mac Mini • mSATA

This thread originally began as an examination of the low cost SATA bridge adapters and “Naked” 2.5” SSDs in the G4 Mac Mini… compared back then with the higher cost (in 2021) of the mSATA and even some Marvell-based adapters and adapter sleds. This now includes some mSATA SSDs on other adapter sleds (like Ableconn, DeLock, etc.) and JMicron-based “White Case” mSATA adapter cards… as costs have now decreased for the mSATAs and some adapters / sleds.

No M.2 SSDs or their adapters were tested here. So unless you’re planning on possibly removing such M.2 SSDs later for use in some newer machine and thus “future proofing” them (as IIO previously attempted to intimate here) IMHO, I don’t really see the need, nor any real advantage of having an M.2 in a G4 Mac Mini over mSATA. I could of course be mistaken.

   

M.2 result on the left, Zheino 128 GB mSATA on the right. I did change the “Expected Peak” setting for the Zheino test (why it only appears to plot lower) and sometimes SCSI Benchmark will auto-set this too. But… compare the Peak & Sustained Read numbers. The Toshiba 24 (ssp3’s) does plot much smoother Reads & Writes… while the Peak and Sustained Write numbers of the Zheino are only just fractionally better.

Here’s the DogFish and Ableconn. Plots a bit more erratic than the Zheino but the “QuickBench 2.0 factored averaging scores” (later here, below) does give it a slight edge over the Zheino on the Ableconn sled. I assume that ATTO SCSI Benchmark’s smoother plotting M.2’s (and mSata’s) might be slightly more expensive?

   

I still prefer the arduous & extensive number crunching approach of QuickBench 2.0 "result averaging" to one-number rate & compare overall configurations and performance.

[*OCD and me.]


HEAT
The Prequel

       
                                                                           It was cool here upstairs this morning with the A/C off and the patio door open.

*Ambient room temp in testing environment here (basement) is a stable 76˚F.

Considering possible operating temperature concerns with the bridge adapters and the “white case” mSata adapters and adapter sleds: have not yet encountered any overheating or extreme heat problems within the G4 Mini. In fact, other than a few small, initial temperature rises on some of the components, or on specific areas of the naked “White Case” adapters… nothing on sleds or adapters has gotten warmer than 112˚ F / after 15-30 minutes runtime. More intense heat could be more prevalent and more of a critical issue in cramped internal-space laptops?

If your G4 Mac Mini’s exhaust temps are above 100˚F very soon after boot, check your fan’s run condition and possibly renew the CPU thermal paste, posthaste.

Initially, all tested Mac mini fan exhaust temps (mini case covers off) here reached 84˚ (maximum) soon after boot and remained there steady. With the exception of two Minis here that quickly went to just above 100˚F almost immediately. [There’s a BIG clue.] Renewed the CPU thermal paste and those temps then matched the 84˚. (And all with no White Case adapter covers in place ~ “naked”.)

Also Heat related: Interesting mSATA flash memory chip up/down orientation discovery… later.

Please note: No Minis here rest solely upon their rubber padded bottoms. All have either small felt or rubber feet attached at each corner to raise them at least 1/8” to facilitate improved heat dispersion and also promote cooler air circulation at (and from around) the base. Consider this a cooling “air gap”. [Much more on HEAT, later.]

Temperatures measured from the center of the inset Apple logo on the base rubber pad of the mini… after 30 minutes run time and with mini’s top case cover in place = 100˚. That’s nothing.

All recent testing and results here - via a 1.42 GHz / G4 Mac mini with 1 GB RAM.


Good News / Bad News?

       

The Ableconn sled with mSATA did exceed the basic SSD + bridge adapters’ overall performance - and for an estimated total combined cost… of now roughly $60.00. Then the “White Case” adapters + mSATA combos come in @around $40.00. While the lowly JMicron - based adapter bridges + SSDs… still hover right around $30.00 total. *All ball park estimates, including the cost of a 120 / 128 GB 2.5” SSD and/or inexpensive mSATA.


MORE  NUMBERS:
Higher numbers are better.

   

Some here (with better mental recall than me) may remember lesser performance numbers reported previously for some of the above, back in August. http://macos9lives.com/smforum/index.php?action=dlattach;topic=5955.0;attach=12611;image Direct consequence of: (1). Using Tiger via a Quicksilver to format and install, (2). Testing in a 1.25 GHz Mac mini, and (3).Utter and complete stupidity. Hurry up and f•ck up!

 “Came to my senses” and realized my folly… just after all those initial test results were all lost -
  when the USB stick they were stored on, failed. So…Using Drive Setup 2.1 definitely matters!


AND NOW… HEAT!

Imagine yourself inside a small enclosed box. Say, with an internal temperature of often greater than +100˚ F. And there’s only one small fan drawing cooler outside air in and around you. And that air is then exhausted through a rather small exit opening. After a prolonged period of time and rising heat, would you: (1). Begin removing articles of clothing to remain cooler, or (2). Put on a coat or some other covering like plastic, steel, aluminum or copper?

   Q: Why would anyone remove plastic covers from mSata adapters (and why are MDD copper heatsinks so highly sought)?
   A: Ther-mal Con-duc-tiv-ity.

       
                     *Higher conductivity numbers are better. (Why shocked about chrome-plated-steel-covered RAM?)

Based (not entirely) on the above I don’t use mSata white case covers, just as I also remove plastic cases from 2.5” SSDs… and I’m now also considering the exclusion of the steel framework of the Ableconn sled in the mini. After all, it isn’t like it’s a laptop to be dragged around bumped & jostled about @ Starbucks.😬



One might also note that the Ableconn mounted mSata faces down towards the mini’s mobo when installed. Which makes it nigh impossible to take open case (booted) temperature readings of the various onboard components. So such temperatures can’t be reported here. And I’ve yet to acquire any of those small, wired, digital probe, temperature monitors. Anyway, monitoring the mini’s fan exhaust temp seems sufficient enough for me at present with the Ableconn. Until I eventually (gasp) buy some of those digital probe temp monitors.

Here are some open case temps measured from the half-height and full-height white case adapters…

 

Above, note the slightly higher temps recorded on the Zheino mSata (compared to those on the DogFish mSata, below). Maybe this is really the (not-so-hot) “hot sandwich” result of the JMicron 20330 being located directly below the mSata mounted on this adapter? Still however, 96˚ F is not really all that hot. The 100.4˚ noted on C19 (lower left, bottom) was a bit surprising. AND a very quick removal of the mSata from this adapter afterwards, produced a temp reading of 96.6˚ from the surface of the JM20330. Again, not really all that hot inside (but closer-quarter laptops might be a different story altogether). The fan’s exhaust temperature was steady @ 84˚.

Just now ran it all again with CD drive installed and the mini’s case cover in place. And in the half-hour or so that it took to run all tests in MacBench 3.0 - the fan exhaust temp rose only to 92.6˚ (from the previous 84˚) and the bottom of the mini’s temperature (center of the logo - rubber pad bottom) read 100˚. The two mSata flash memory chips rose to 98.6˚ and the two round capacitors rose to 99.7˚. D1 and D2 read 97.5˚ and 95.8˚.

Still, temperature measurements taken while under a much heavier CPU load would also be useful - but for now there’s at least +36 F degrees to go, before I’d ever even really become concerned.

You might also note the “up” orientation of the flash chips on the Zheino mSata pictured above - compared to the “down” orientation of the flash chips on the DogFish mSata below. Perhaps something to consider when selecting mSata’s and mSata adapters? Up? Or down in relation to the mounting socket? Think I’ll leave the Zheino on the “half-height” card above as the flash mem chips don’t face down towards the JM20330 chip, whereas the DogFish mSata would place the surface of its’ flash chips nearer the JM20330 on the half-height. So perhaps I’ll “err” more towards the dreaded (not-so-hot) “heat sandwich” in the Mac mini with the half-height card as my own personal choice. (ala NewEgg.com)

But it seems to run hotter! (I don’t care.)


                                                                              There’s a difference? Or are all mSata flash chips oriented just one way now?



The “full-height” adapter card (above / from Amazon) seems to run cooler in the open tests. (Mac mini & plastic cover removed.) However, the white-circled trace area reads 98˚… while no traces on the half-height card approximate that. This may likely be the better / more cool-er running adapter due to the lower temps in an open mini but I am concerned about its’ overall length as it may restrict air flow. The closer such things extend to near the front of the mini, the more they are also exposed to mobo components that do generate more heat and the longer length might trap more of that heat under (and absorb into) the full-height card and its’ components. Time will tell.

Also, the vertical red line in the above image might be where I’d consider removing a part of the PCB to allow for more air flow. But no, I won’t be hacking up my Ableconn… I’ll just leave the steel mounting framework off. The Ableconn with the metal mounting frame removed is only 2.25 inches wide and this provides an eighth-inch more open space around either side for air circulation. And again… the Ableconn sled + DogFish mSata scored best overall performance in tests here.

Any way the wind blows…?


SPACE

Here are some grab shots of the space that each approach occupies inside the mini (sans CD drive). I personally prefer those that take up less actual space, thinking that these may increase the (sic) “Dyson Cyclonic”  Vortex™ air circulation pattern with that added space and thereby increase the likelihood of possibly better airflow, and thus better cooling? It’s a theory. Really should re-install an original HD and check fan exhaust and bottom temps for comparison too. Hotter, or not? (Did it. Not really hotter.)

Red outlined area in this first image encompasses a BIG heat generating area of the mini’s mobo.

 

 

 

 


SHUT UP ALREADY!
And other errata…

Caveat: This is by no means intended as an all-inclusive or Best-of-the-Best representation of all the available mSATAs, or of all the various available mSATA adapters / nor of all SSDs and other IDE bridge adapters or upgrade routes. There are certainly many more mSATAs, SSDs and various other adapters… and most certainly more expensive choices than those represented & tested here. But hopefully, this will serve as a broad sampling of some of the lower… to somewhat higher-cost possible choices.

As always… with tested performance versus cost = Value. If you’ve a big budget and want the fastest, best-est, etc… well, it’s your money.

And as merely one of many possible “for instances” or “what-abouts” - consider the following bit of info concerning some differences between a Zheino mSata type M3 and type Q3:

“Flash memory chips used on the Q3 are from Intel, while the M3 uses flash memory chips from Sandisk. Of these two, the M3 appears to be a little faster reading, but the Q3 is a little faster at writing. However, both M3 and Q3 use a DRAM-less design, which means that under sustained write load (such as when restoring this SSD from a backup), the write rate is initially fast, but after some time, it will eventually drop to around 28MB/s for the M3 model, and around 38MB/s for the Q3 model. This is slower than MLC mSATA SSDs, which will sustain write speeds of 110MB/s from start to end. However, such long writes are uncommon in normal usage, so therefore it should not matter quite as much. This slow down of the write throughput should also be less pronounced if there are pauses in between the writes, and if there is some unallocated space for over-provisioning. So in conclusion, the M3 is a decent performer, it disappoints a bit in terms of sustained write rates, but it's still a good value, I think.”  -paraphrased and quoted from a review by S.Lin “Good Performance and Value”

The DogFish mSata tested here does have those Sandisk flash chips.
So there’s a myriad of variables and other minutiae that one might consider when choosing any mSata / SSD application and installation configuration(s)?


“Stupid is as stupid does.”

Wait, you’ve already heard this tale of woe.

Initially began all of this testing by fresh formatting the mSata’s and the stripped 2.5 SSDs in a Quicksilver using Tiger’s Disc Utility. So after the first full week of testing and benchmarking all various installations in the full range of G4 Mac minis (from the 1.25 GHz to the 1.5 GHz models) with QuickBench 2.0 and ATTO’s SCSI Benchmark… ALL of those benchmarks were less than impressive. And luckily all of those tests, results and companion / comparison graphics were lost when the USB stick I was using to transfer and store all of that (shuffled between upstairs and downstairs machines) failed utterly. And testing began all over again (only with the 1.42 GHz model this time) after I remembered my own “best approach” of formatting with Drive Setup 2.1 for optimal performance from OS 9.2 in the Mac mini. USE DRIVE SETUP 2.1!!! I cannot stress this enough.

However… supporting benchmarks and other graphics do now exist here for all the above test results. BUT as lengthy and verbose as this already is - they are not included here. (Collective sigh of relief.)


Seasonal Lagniappe:

And sorry (not sorry) but while this IS about G4 Mac minis… and it is October.
HAD to share these images of a mini recently acquired by one of our more infamous cohorts.
The thing reportedly booted right up and in this condition!

Presenting the “Halloween Edition” G4 Mac mini. Very scary.

 

 

 
                                                                                                                                 Very nice spider und web!
“We’re often surprised by things that shouldn’t work (but do) and by things that should work (but do not).
 Of course we all appreciate more, the former… rather than the latter.”

доверяй, но проверяй! 

[DieHard]

FBz that was an awesome post and I will be re-reading it for a 3rd and forth go in the morning.

Yeah... full disclosure to members, those were my pics of that very abused mini that was shipped to me from the member known as Redicliff.

He contributed that G4 mini spider apartment (and some ministacks) to the cause and I thank him !  And yes, it booted to OS X without issue, lol.

The mini now looks like brand new and I really should have posted some after pics, the spider was long since dead so no insects were harmed in this production... getting the CD to function was quite the challenge.

[ssp3]

I see a couple of errors and misconceptions, but so be it.