Saturday, January 20th 2024

SSD Overclocking? It can be Done, with Serious Performance Gains

The PC master race has yielded many interesting activities for enthusiasts alike, with perhaps the pinnacle of activities being overclocking. Usually, subjects for overclocking include CPUs, GPUs, and RAM, with other components not actually being capable of overclocking. However, the enthusiast force never seems to settle, and today, we have proof of overclocking an off-the-shelf 2.5-inch SATA III NAND Flash SSD thanks to Gabriel Ferraz, a Computer Engineering graduate, and TechPowerUp's SSD database maintainer. He uses the RZX Pro 256 GB SSD in the video, a generic NAND Flash drive. The RZX Pro uses the Silicon Motion SM2259XT2 single-core, 32-bit ARC CPU running up to 550 MHz. It has two channels at 400 MHz, each with eight chip enable interconnects, allowing up to 16 NAND Flash dies to operate. The SSD doesn't feature a DRAM cache or support a host memory buffer. It has only one NAND Flash memory chip from Kioxia, uses BiCS FLASH 4 architecture, has 96 layers, and has 256 GB capacity.

While this NAND Flash die is rated for up to 400 MHz or 800 MT/s, it only ran at less than half the speed at 193.75 MHz or 387.5 MT/s at default settings. Gabriel acquired a SATA III to USB 3.0 adapter with a JMS578 bridge chip to perform the overclock. This adapter allows hot swapping of SSDs without the need to turn off the PC. He shorted two terminals in the drive's PCB to get the SSD to operate without its default safe mode. Mass Production Tools (MPTools), which OEMs use to flash SSDs, were used to change the firmware settings. Each NAND Flash architecture has its own special version of MPTools. The software directly shows control of the Flash clock, CPU clock, and output driving. However, additional tweaks like Flash IO driving with subdivisions need modifications. Control and Flash On-Die Termination (ODT) and Schmitt window trigger (referring to the Schmitt trigger comparator circuit) also needed a few modifications to make it work.
The overclock results were an increase of 17.6% for the Silicon Motion controller, up to 500 MHz from the original 400 MHz, and 106% with 400 MHz clock for the NAND Flash, which only operated at 193 MHz stock. The performance results? While it varies from benchmark to benchmark, the latency has dropped. Only a slight improvement has been made in reading speeds, while writing has seen a more significant impact. Bandwidth has been increased, so the benchmarks naturally responded well. The biggest problem is the increased operating temperature. Temps were stable at 40C in stock settings, while the overclock pushed it to 45C. Over the long run, this decreases the lifetime of the drive. Overclocking the SSD also voids all manufacturer-implied warranties and reduces the expected number of read/write operations.
Below, you can see a few benchmarks by Gabriel Ferraz, while the rest are available in the YouTube video here.
Source: Gabriel Ferraz (YouTube)
Add your own comment

41 Comments on SSD Overclocking? It can be Done, with Serious Performance Gains

#26
delshay
Does anyone know how to read the buffered memory timings inside an SSD?

They normally come with 256, 512, 1MB ect, ect depending on size of SSD. ...Example: I have 2TB SSD & I know it has 2MB DDR2 memory onboard. But is there a way to read it's memory timings,
Posted on Reply
#27
mechtech
"While this NAND Flash die is rated for up to 400 MHz or 800 MT/s, it only ran at less than half the speed at 193.75 MHz or 387.5 MT/s at default settings."

If it's rate for 400MHz, why run it at less than half?? I mean if you paid for those chips why not run them closer or at rated speed and sell the product at a higher price since it has higher performance??
Posted on Reply
#28
Count von Schwalbe
mechtech"While this NAND Flash die is rated for up to 400 MHz or 800 MT/s, it only ran at less than half the speed at 193.75 MHz or 387.5 MT/s at default settings."

If it's rate for 400MHz, why run it at less than half?? I mean if you paid for those chips why not run them closer or at rated speed and sell the product at a higher price since it has higher performance??
Bottlenecks elsewhere, such as the controller. If you don't have the bandwidth you may as well save the power.
Posted on Reply
#29
R-T-B
QuitessaIf you only unlock the full 400MHz of the nand I'd say it's less of an overclock and more of a true performance unlock, or a de-hobbling.
He went to 500MHz.
ThrashZoneHi,
Sure isn't that called mem cache :cool:
No, this is something else.
Posted on Reply
#30
Bwaze
"Usually, subjects for overclocking include CPUs, GPUs, and RAM, with other components not actually being capable of overclocking."

Au contraire, almost nothing is overclockable these days to any useful degree. SSDs included.
Posted on Reply
#31
Broken Processor
Can't help but think this is a terrible idea, definitely cool but no just don't do it on you home putter.
Posted on Reply
#32
Sabotaged_Enigma
Finally something interesting and creative appeared in the PC industry! I haven't been reading in a lot of curiosity for years! Good job!
Posted on Reply
#33
Count von Schwalbe
R-T-BHe went to 500MHz.
On the controller. Flash was 400 MHz.
AleksandarKThe overclock results were an increase of 17.6% for the Silicon Motion controller, up to 500 MHz from the original 400 MHz, and 106% with 400 MHz clock for the NAND Flash, which only operated at 193 MHz stock.
Posted on Reply
#34
A&P211
Can this overclock my girlfriend to teach her about how to turn on a PC, really she needs to learn about PC basics.
Posted on Reply
#35
hojnikb
JismI am curious tho about those tools used. I have a old Intel 80GB ssd that i would love to toy around with.

Even 10% increase for such drives would be huge.
These are so called Mass Production Tools (MPTools for short). You won't find any kinds of tools for 1st party controller vendors like Intel or Samsung, because they don't sell their controllers to other vendors. So unless someone leaks them from their factory, it's a fat chance you'll get one.

SiliconMotion, Phison, Maxiotek and others usually sell controllers and/or complete board designs to vendors, which they can customize using said MPTools. Because many factories/vendors then manufacture these drives, tools like that usually quickly leak to the interwebs.

If anyone is brave enough, there's an awesome resource for such tools. It's in RU, but Google translate fixes that. Beware, that wrong settings can easily brick your drive.
www.usbdev.ru/files/
lexluthermiesterSince no one seems to mention it, I will: What does this do to NAND durability?

Overclocking other types of electronics is doable because we're generally talking about transistors or other very durable components. NAND flash cells are NOT durable. They are in fact relatively very delicate. So overclocking them? Seems a bit dubious to me. Granted, the article and whatnot above talks about the NAND controller as much as the NAND itself, but still.
Likely not much. This is NAND interface overclock (ie the bus connecting the nand die to the flash controller -- similar to SATA). At worst, you'd get data curruption, because of errors on the interface itself.

I'm guessing bulk of the performance gain is from the controller overclock, not the interface. It's likely factory chose that toggleddr frequency for a reason (ie. die itself isn't capable of higher speeds, so faster interface clocks aren't needed).

Might need further testing.
Posted on Reply
#36
lexluthermiester
hojnikbLikely not much. This is NAND interface overclock (ie the bus connecting the nand die to the flash controller -- similar to SATA). At worst, you'd get data curruption, because of errors on the interface itself.
You might be misunderstanding how NAND works. Overclocking NAND is iffy at best.
hojnikbI'm guessing bulk of the performance gain is from the controller overclock, not the interface. It's likely factory chose that toggleddr frequency for a reason (ie. die itself isn't capable of higher speeds, so faster interface clocks aren't needed).
Plausible. But as you said...
hojnikbMight need further testing.
... further testing is needed.
Posted on Reply
#37
GabrielLP14
SSD DB Maintainer
On the contrary, what really helped out was the NAND clock increase, that was the first and main reasons for the performance uplift
Posted on Reply
#38
lexluthermiester
GabrielLP14On the contrary, what really helped out was the NAND clock increase, that was the first and main reasons for the performance uplift
I'm not arguing that point. I'm arguing the wisdom of doing it at all. NAND is NOT durable. It degrades over time under normal use, to say nothing about over clocking which very likely degrades durability greatly.
Posted on Reply
#39
Count von Schwalbe
lexluthermiesterI'm not arguing that point. I'm arguing the wisdom of doing it at all. NAND is NOT durable. It degrades over time under normal use, to say nothing about over clocking which very likely degrades durability greatly.
The NAND die was rated for the 400MHz clock.
Posted on Reply
#40
GabrielLP14
SSD DB Maintainer
lexluthermiesterI'm not arguing that point. I'm arguing the wisdom of doing it at all. NAND is NOT durable. It degrades over time under normal use, to say nothing about over clocking which very likely degrades durability greatly.
Like the own video title, its just a curiosity, not something i recommend to do, sine as you can see in the video, has a high chance of things go wrong.
Count von SchwalbeThe NAND die was rated for the 400MHz clock.
originaly yes, but like i mentioned, there are many reasons why they downclocked it
Posted on Reply
#41
hojnikb
GabrielLP14On the contrary, what really helped out was the NAND clock increase, that was the first and main reasons for the performance uplift
That's really interesting. I really do think that ToggleDDR (what you're overclocking) is the interface between flash controller and the raw nand die itself. (sort of like sata between spinning rust and motherboard).

Overclocking just the interface shouldn't affect nand performance, unless the interface speed is lower than what nand die is actually capable of delivering.

I might be wrong, i'd love to get some better insight about this.
Posted on Reply
Add your own comment
Nov 21st, 2024 11:09 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts