First there is a mountain, then there is no mountain, then there is….Or, once more unto the breach.It’s easy to cast “horrible” and “junk” aspersions on
all JMicron / JM20330 based adapters, especially if you’ve never tested an
EVB-002-3 “Bribge” adapter. And it is likely that many here have had their share of JMicron based adapters that do not work, or work poorly - and testimony on many of those clunkers is scattered throughout this Forum. (The 44-pin, MCA-004 V1.3 immediately and painfully comes to my mind.) But that’s another "G4 Mac mini" story.
http://macos9lives.com/smforum/index.php/topic,5955.msg44360.html#msg44360However, it is very hard to beat a Bribge EVB-002-3 for its’ low cost and
near-matched performance up against Marvell based adapters… especially in single SSD and some dual drive situations. All of the Bribges here are tried, tested and true. Even if they can’t spell ‘Bridge’ or ‘Technology’.
“But it doesn’t cost very much. It simply can’t be as good as the more expensive Marvell based adapters.”
Benchmark Methodology…Not all just about the Bribges.
The gnat’s ass approach.Most of us are familiar with the
Numeric
View panel of QuickBench 1.5 and 2.0. And most of us will run only that first
NV panel and call that good. Now tell me that just looking at all those rows of numbers that you can really get a good assessment of your tested drive’s overall performance.
By just looking at the Blue “Peak Performance” numbers? And some will even run the Extended panel too. But still, what do you really know that you can readily compare to another drive or any modification(s) to your machine(s)?
Bupkis.
Well this blue number IS bigger than that blue number.Now maybe everyone else is already doing this and I just woke up to the fact that if you add all of those rows of numbers together and then average your results for each column, and then add those totals together… and then divide by the number of columns - you will arrive a single “averaged” number score for that panel. Too much? It only gets worse.
Then if you’ve also ran the Extended panel (and the Extended “Plus”/ 20-100 MB panel in QuickBench 2.0) you may have two more panels to add, average and then add those results all together AND average all
that to arrive at a single, solitary, all-inclusive, overall performance number. [Viola!] That, you can then compare to other drives or modification states. I’ve been calling this OPC. You’ll see all this later, below. But again, it was only me that wasn’t already doing all of this, right?
And the real trick of this… the glimmering
“eye and mind saving” alternative… is
QuickBench 4.0. because it adds and averages ALL of the columns in all of the panels for you. (But it only runs under OS X 10.2 or above.) You still must add and average a few numbers, but nothing like the longhand spreadsheet example below of this QuickBench 1.5 example. Hallelujah indeed! All results here are taken from bootable OS 9.2.2 partitions.
Remember, the first two lines of the Numeric View are not used. Remember the first two lines of…
The first two lines in the Numeric View from QB 1.5 and 2.0 are excluded in order to match the sample ranges tested and reported under QB 4.0. I tested and compared both and results are nearly identical, if not exactly. Those 1st two lines often contain KB/sec results and I’m to lazy to move decimal points anyway to convert them to fractions of a MB.
If I’ve not completely lost you already, you might see that the OPC for this March 2022, QuickBench 1.5 benchmark test of a Seagate 60 GB 5400 RPM HDD in a B&W will yield you
19.20. That’s
15.766404 plus
22.648444 =
38.414844. Divide that by the TWO panels, to then arrive at the
19.20 overall averaged score.
AND then that… subtracted from the
25.74 (noted in Bribge tests further down below here) will provide a difference of +5.86 points over the Seagate HDD. So roughly, the SSD + Bribge nets you about a
34% increase in performance over the Seagate -
if my math is correct. Mind you, this is only a two-panel averaged score from QuickBench 1.5 / and three test panels are better & more accurate. But this two-panel is intended here only as an easy example.
So based on the above, if you were able to acquire a Bribge here from the Forum for $3.99 (only in the Continental U.S.) and you can pick up a low cost 128 GB SSD (like the Inland brand SSD from MicroCenter last week - $14.99) you could increase your B&W’s performance by 34%… for right around
$20.00. Or... for less than what a StarTech Marvell-based adapter would cost you alone. And
that StarTech…
without a 128 GB SSD.
A lot to absorb here above - but I had to get it all out of my head. Mea culpa. / Clear-As-Mud.
More Benchmarking!Way less exciting than a cowbell.Here are the benchmark comparisons from a B&W G3, Rev.2. 450 MHz.
I know, not the same as a Rev.1 but I’m not swapping mobos (yet.)
Two-panel QuickBench 2.0 results here:
Most seem content to only run the first “
Numeric
View” panel and only compare those simple results.
Above, w/ QuickBench 2.0 - included Extended Panel for more complete spectrum comparisons.
Numeric only: 20 vs. 34 vs. 44. Or
two test panels combined results:
25 vs.
47 vs.
67?
Seems apparent that a flashed SiL3112 card
can nearly double performance.
But then… the Sonnet Tempo PCI card in the B&W video slot - scores
67.
[For feeling that need… for feeling that speed.] 
Above, also with QB 2.0 (and
4.0) and with the 20-100 MB panel results also included. Much better?
And why QB 4.0? Each test panel auto-averaged by 4.0 / negates need to manual add 2.0 columns.
And including the 20-100 MB panel provides a much more inclusive OA performance assessment.
Enter The Bribge(Above was all StarTech baby.) 
Tested more than once to confirm these results.
Above… the most recent head-to-head comparisons between Bribge
and a StarTech in a B&W G3
Rev.2 / using
QuickBench 4.0.
Now again,
QuickBench 4.0 reportedly will not run on anything less than OS 10.2 - so it was used here from a booted Tiger partition / and on two very different OS 9 partitions… and with very nearly identical results. After many recent weeks of benchmarking here -
QB 4.0 has now been adopted here when it can be used in this manner. (And no, not just on the Bribge and B&W… on MDDs, Quicksilvers, Digital Audios, etc.)
On older machines (or those incapable of running OS 10.2 or above) one may sadly still be adding all of those
QuickBench 2.0 columns of numbers. (Thanks to Borgmac for the QuickBench 4.0 gentle nudge.) If only I had known.
Do not expand this image for larger view. *Also: 1 KByte and 2 KByte QB 2.0
Numeric View results omitted for comparison QuickBench 4.0’s test sampling.
Many thanks to Ms. D. Lenoir for her unwavering math skills and patience.
QuickBench 2.0 on the left, QuickBench 4.0 on the right.
So… should anyone spend more money for a Marvell based adapter in a B&W Rev.1 (or even a flashed SiL3112 SATA Controller card)? I wouldn’t. Not at least until I could get something less expensive (other than the original B&W’s HDD) to sucessfully boot from the Rev.1 mobo. Not so quick to “throw money” at these vintage Blue and White machines. And isn’t the primary focus “cost vs. performance” anyway?
I’ll be re-installing a Rev.1 mobo in one of the B&Ws here. But I am
almost certain that I used a Bribge in one of the B&W Rev.1’s here - before I eventually converted it to a Rev.2 mobo. Do keep reading if you’ve made it this far. “Gripping” I know.
AND NOW THE B&W - REV.1Yes, I finally had to do the swap.
I don’t know what it is about the B&W G3s over on that “other side of the pond” but a pattern seems to be forming? Maybe it’s the difference in electrical voltage? I haven’t a clue.
See: http://macos9lives.com/smforum/index.php/topic,6225.0.html The B&Ws remain cantankerous where-ever they’re plugged-in. Rev.1 or Rev.2 mobos.
So, swapped out a B&W Rev.2 with a Rev.1 mobo and tested
everything all over again.
And not only just with Tiger. Might also notice the CPU swapped from 450 MHz to 350 MHz?
Performance? Feeling the feel.Cold boot times with OS 9.2.2 ranged from 28.65 seconds to 33.87 with the Bribge.
And 40.98 seconds to 41.35 seconds with OS 10.4.11 (Tiger) also w/ Bribge.
That’s total time elapsed from button-push to a complete desktop display.
With the StarTech it was 31.22 seconds for OS 9 and 50.86 to 51.18 seconds for Tiger.
(Scout’s honor!)
Now also keep in mind - less than 1 GB of total RAM installed here… but close.
Why would anyone pay more for a StarTech in a B&W G3 (Rev.1 or 2)?
OR… for use in G4s (other than the Cable Select required MDDs)?
But, you can also use Bribges in the MDDs (set to Master) in MDD single drive scenarios.
As for G40’s B&W Rev.1 problem… yes it could be the Bribge.
BUT
it works in his Digital Audio! AND it works here in a Rev.1.
Someone in the U.K. or the EU needs to report a working Bribge
in a B&W (Rev.1). IIO is that you? Anybody? It's not the Bribge!
Believe that G40 is now opting to leave the Hynix + Bribge
in his DA and returning an HDD back into his B&W Rev.1.
“…although the G3 B&W undoubtedly has style, I am going to recommend
newbies skip this model and grab a DA, QS, or MDD G4.” *Out of context here - but most certainly a very apt point.Bribges (EVB-002-3) in the Continental U.S. here:
http://macos9lives.com/smforum/index.php/topic,5967.0.htmlHopefully, this finally concludes the
once-and-for-frikken-all StarTech-Marvell vs. Bribge-JMicron misCONception.
But wait… there’s more!And now in closing (yes really/ finally)… from Petros90:
http://macos9lives.com/smforum/index.php/topic,6225.msg46669.html#msg46669“So why don't we just cut out the middleman, and replace the OS9 on the SSD with DieHard's excellent final Last OS9?
Procedure - boot from CD, copy the original SSD OS9 System Folder onto the Panther partition, de-bless just in case, copy the DieHard OS9.2.2 System Folder to the OS9 partition, bless, reboot. The whole thing took minutes.
Now boots from any OS to any OS. The only caveat is that on booting into OS9 a dialog box appears asking for the Install CD.”
@G40: You might try the above with the Bribge + Hynix in the DA and then move it back into the B&W. I assume that Petros90 only copied the SSD’s OS 9 System folder to the Panther partition (in your case, the Tiger partition?) in order to save it… should he need to move it back to its’ original location afterwards. AND Petros90’s B&W also used a Rev.2 mobo.
At this point, who knows? It could work. Maybe? Always just one more thing. 😉
Nah, I’m finished.Sorry, no pretty pictures… Well, maybe a
prettier one?
Here’s a QuickBench 4.0 result from an SSD in a Quicksilver, destined for eventual use in an MDD.
Maybe its’ recipient will report on a better than
55.84 score, once it is MDD installed and tested?
Click on the image, it’s 15 inches wide full-size. And it’s “just a bit better” than the B&W’s
25.74.
Topic: B&W G3 unhappy with SSD. (Read 3412 times)