Jump to content

Recommended Posts

Posted

PATA to SATA Adapter Speed Test and Bottleneck Assessment

October 19, 2024

Joel Peterson / Custom Xbox Direct

Full PDF with technical documents available here:

PDF Google Drive Link

 

Introduction:

With the decreasing lifespan of original Xbox hard disk drives and recent advancements in BIOS modifications allowing for increased Ultra Direct Memory Access (UDMA) modes, the need for a fast and reliable device to convert the Xbox’s legacy PATA (IDE) interface to a modern SATA one has become an important point of interest, discussion and discourse.

What follows is an assessment of 2 currently available adapters, 1 discontinued adapter and 1 adapter currently in development. Testing was performed in a real-world Xbox ecosystem as well as a synthetic benchmark scenario intended to discern if any of the adapters themselves would act as a transfer speed bottleneck.  In all situations great care was taken to ensure that the only variable being altered per test was the adapter itself.

Adapters: Technical details regarding adapters and integrated circuits (IC) are available at the end of this report

The controller chips on generic green PCB adapters (DOF indicator) are not true JM20330 chips. They are clones produced by a Chinese company, however, testing indicates that any design concessions that were made, if any, do not impact performance in a major way from the original JM20330 specifications. Both the Marvell 88SA8052 and the JM20330 advertise maximum PATA speeds of 150MB/s on their technical sheets. The MO3E adapter MUST be modified with the removal of the 0ohm resistor at R5, a 100ohm resistor added at R4 and a 10Kohm resistor added at RH1. The device will perform at roughly 50% speed if this modification is not performed.

Startech Marvell 88SA8052-NNC2 PATA to SATA Bridge (Currently Available)

spacer.png

Xenium Prototype Marvell 88SA8052-TBC2 PATA to SATA Bridge (Under Development)

spacer.png

Generic Adapter “2036” JM20330 Clone PATA to SATA Bridge (Discontinued, Old Stock Available)

lbcHEKE.png

Resistor Modified “MO3E” JM20330 Clone PATA to SATA Bridge (Currently Available)

spacer.png

spacer.png

Results:

Both synthetic and real-world test were performed with the same Samsung 870 EVO 500GB SSD.

Test Scenario 1

Synthetic Benchmarks Attempting to Saturate Adapter Throughput at UDMA-5

UDMA-5 is the maximum mode supported by the motherboard used in the following tests

spacer.png

spacer.png

 

Repeatable results at higher file size transfers indicate that bottleneck has been achieved in all 4 adapters. The similarities between the Marvell IC based adapters and the differences between them and the JM IC based adapters lends support to the theory that the test was successful in showing the maximum performance achievable by each adapter in a UDMA-5 defined environment.

Test Scenario 2

Boot Speed Benchmarks Attempting to Demonstrate Real-World Behavior

UDMA-6 is the maximum mode supported by the motherboard used in the following tests.

spacer.png

Full video available here: https://youtu.be/c57VuqYLG3A

spacer.png

Dashboard Cold Power Cycled Boot Time in Seconds                    Lower Time Number is Better

Key:

UDMA6 Config: Cerbios is configured for UDMA-2 before flashing (standard configuration)– BUT the Cerbios.ini file present on C drive indicates the use of UDMA-6 instead

UDMA2 M8+TSOP: Legacy modded BIOS is booted from TSOP (Note boot flubber sequence is shorter than Cerbios which results in an overall faster boot time)

UDMA6 BIOS & Config: Cerbios is configured for UDMA-6 before flashing and the Cerbios.ini file present on C drive also directs the use of UDMA-6

UDMA2 BIOS & Config: Cerbios is configured for UDMA-2 before flashing and the Cerbios.ini file present on C drive also directs the use of UDMA2-2

 

These results indicate that in multiple use case scenarios, there are no major failings by any of the adapters. All adapter results are roughly within 1 second of each other. There are some unexpected curiosities such as the 1 full second decrease in boot time with the Startech adapter with the Cerbios config file present versus without it. More interestingly, the Generic MO3E adapter shows exceptional performance in a Cerbios UDMA2 configuration, beating the Startech by 1 second. Each of the time trials were conducted 3 times from a cold power cycled boot with consistently repeatable results within +/- 0:033 accuracy per unit trial.

 

Conclusion

In UDMA-5 synthetic tests, adapters based on the Marvell 88SA8052 IC show data read speeds 7.57% faster than their generic clone JM20330 counterparts in a maximum transfer bottleneck scenario. This percentage is based on the average of all adapters within their respective IC.

In UDMA-6 boot speed tests, the picture is a bit less concise with the Marvell 88SA8052 only booting 1.17% faster than the JM20330 devices in the BIOS & Config scenario and with there being no measurable difference in a config file only dictated scenario.

 

Filtering the results for the only currently available adapters yields a clearer differential.

UDMA-5 synthetic Bottleneck Throughput

Startech is 8.29% faster than the modded MO3E

 

UDMA-6 Real-World Xbox Cold Boot Speed

Startech is 0.00% faster than the modded MO3E +/- 0:033 seconds

 

UDMA-2 Real-World Xbox Cold Boot Speed

Startech is 0.08% faster than the modded MO3E +/- 0:033 seconds

 

 

Caveats

Price in USD

Startech: $18.99

MO3E (unmodded): $2.77

MO3E (modded by servicer): $11.99 - $15.00

Generic adapters such as the MO3E and 2036 have had a history of poor build quality, missing parts and inconsistent soldering. However, with the most recent release of the MO3E/MO3C in mid-2024, the factory quality has improved greatly. The only problem of course is that the resistors need to be reconfigured to achieve proper speed and communication with the Xbox motherboard.

 

Final Takeaway

The Startech adapter and others based on the Marvell 88SA8052 IC are considered to be more advanced and premium by the general consumer, which is backed up by excellent build quality, an industry name brand and tangible speed improvements in synthetic testing versus the cloned JM20330. However, real-world applications within the Xbox service and modding community show these benefits to be mostly unnecessary and superfluous. The Startech adapter shows no measurable improvements in boot times in either UDMA-6 or UDMA-2 scenarios versus the generic JM20330 clone.

 

Special Thanks to Chris Grandin and Xenium Nemesis

PATA to SATA Adapter Speed Test and Bottleneck Assessment 2024.pdf

  • Like 1

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...