Storage


Exploring Performance Consistency in USB 3.0 Flash Drives

Exploring Performance Consistency in USB 3.0 Flash Drives

SSD technology has advanced to a stage where it is possible to have flash drives with bona-fide SSD controllers. The addition of a SATA to USB 3.0 bridge provides an ubiquitous high-speed direct attached storage interface. We have already looked at the Phison S9-based Corsair Flash Voyager GTX and the SandForce SF2281-based Mushkin Ventura Ultra. Our review of the Mushkin Ventura Ultra 120GB drive had a comment about the device’s thermal throttling issues. This piqued our interest and we set out to see if any of the other units evaluated before had such issues. Obviously, we were also interested in determining whether our Ventura Ultra sample had the problem. In the case of throttling, we also wanted to make our review complete by bringing out the circumstances under which it would trigger. As we will see further down in this piece, our experiments did find evidence of throttling in the Mushkin Ventura Ultra.

Recently, we also received a sample of another SSD controller-based flash drive. Putting that drive through our throttling test revealed even worse results compared to the Mushkin Ventura Ultra. Upon our feedback, the manufacturer is contemplating a firmware update. In this piece, we will refer to the unit as ‘BrandX 240GB’ (a complete review of the unit with benchmark numbers will be published once the manufacturer provides us a resolution for the throttling issue). The list of DAS units put under the performance consistency scanner is provided below.

  • Mushkin Ventura Ultra 120GB
  • BrandX 240GB
  • Corsair Voyager GTX 256GB
  • LaCie Rugged Thunderbolt 500GB
  • Mushkin Atom 64GB

Performance Consistency Evaluation Methodology

Our testing methodology for DAS units takes into consideration the usual use-case for such devices. The most common usage scenario is transfer of large amounts of photos and videos to and from the unit. In order to tackle this, we created three test folders with the following characteristics:

  • Photos: 15.6 GB collection of 4320 photos (RAW as well as JPEGs) in 61 sub-folders
  • Videos: 16.1 GB collection of 244 videos (MP4 as well as MOVs) in 6 sub-folders
  • BR: 10.7 GB Blu-ray folder structure of the IDT Benchmark Blu-ray (the same that we use in our robocopy tests for NAS systems)

These folders were put in a RAM drive on our testbed. Using robocopy, each folder was copied from the RAM drive to the DAS (write benchmark) and back to the RAM drive (read benchmark). This process was repeated thrice. To check for performance consistency, we instrumented our benchmark script to record the flash drive’s read and write transfer rates while the robocopy process took place in the background. For supported drives, we also recorded the internal temperature of the drive during the process.

Evaluation Results

Aspects that may influence performance consistency include thermal throttling and firmware caps on access rates to avoid overheating or other similar scenarios. This aspect is an important one, as the last thing that users want to see when copying over, say, 100 GB of data to the flash drive, is the transfer rate going to USB 2.0 speeds. The graphs below show the speeds observed during our real-world DAS suite processing with the instrumented benchmark script. The first three sets of writes and reads correspond to the photos suite. A small gap (for the transfer of the videos suite from the primary drive to the RAM drive) is followed by three sets for the next data set. Another small RAM-drive transfer gap is followed by three sets for the Blu-ray folder.

An important point to note here is that each of the first three blue and green areas correspond to 15.6 GB of writes and reads respectively. Throttling, if any, is apparent within the processing of the photos suite itself.

Mushkin Ventura Ultra 120GBBrandX 240GBCorsair Voyager GTX 256GBLaCie Rugged Thunderbolt 500GBMushkin Atom 64GB

The Mushkin Ventura Ultra 120GB does throttle after the S.M.A.R.T temperature report reaches 45 C. In our testbed (ambient at 22C), this temperature was reached after approximately 18 GB of writes and 16 GB of reads without any interruptions. Beyond this, the write speeds got capped around 80 MBps (compared to the non-throttled 100 MBps), while the reads got capped around 100 MBps (compared to the non-throttled 140 MBps). The throttling allowed the temperature to stay below 50 C throughout our benchmark run.

BrandX, though, behaved differently. The temperature in the steady state settled around 53 C. Unfortunately, the throttling (after a similar amount of data transfer as the Mushkin Ventura Ultra) resulted in the write speeds getting capped at a paltry 20 MBps. The reads managed to remain around 80 MBps.

The Mushkin Atom is not meant to take the performance crown, and the advertised write speeds are in the 20 MBps range. Our benchmark run ran at a steady 15 MBps, while the reads didn’t get throttled at all in our evaluation process.

The LaCie Rugged Thunderbolt drive is not a flash drive (unlike the other units considered in this piece). It is a bona-fide SSD in a USB 3.0 / Thunderbolt hybrid enclosure. It is no wonder that the transfer rates were hardly affected by the relatively small amount of data traffic. The real winner of our evaluation was the Corsair Voyager GTX. There is no performance degradation even after more than 120 GB of continuous traffic. It must also be remembered that we ran our robocopy benchmark with multiple queue depths in our original review (that was more than 480 GB of continuous traffic) and the table showed remarkable consistency. The instrumented benchmark run confirmed the same.

Concluding Remarks

Coming to the business end of the piece, it must be remembered that the capabilities must be compared keeping the price in mind. If cost is not an issue and the form factor is not an issue, the LaCie Rugged Thunderbolt is the hands down winner for a fast portable DAS unit with consistent performance. If cost is the primary concern, and one doesn’t mind a little bit of throttling after around 18 GB of continuous write traffic, the Mushkin Ventura Ultra is a great choice at less than $0.6/GB. The middle ground (right balance of cost and fast, consistent performance, along with a traditional flash drive form factor) is held by the Corsair Voyager GTX. At more than $0.85/GB, it also serves as a reminder that one gets what they pay for.

Exploring Performance Consistency in USB 3.0 Flash Drives

Exploring Performance Consistency in USB 3.0 Flash Drives

SSD technology has advanced to a stage where it is possible to have flash drives with bona-fide SSD controllers. The addition of a SATA to USB 3.0 bridge provides an ubiquitous high-speed direct attached storage interface. We have already looked at the Phison S9-based Corsair Flash Voyager GTX and the SandForce SF2281-based Mushkin Ventura Ultra. Our review of the Mushkin Ventura Ultra 120GB drive had a comment about the device’s thermal throttling issues. This piqued our interest and we set out to see if any of the other units evaluated before had such issues. Obviously, we were also interested in determining whether our Ventura Ultra sample had the problem. In the case of throttling, we also wanted to make our review complete by bringing out the circumstances under which it would trigger. As we will see further down in this piece, our experiments did find evidence of throttling in the Mushkin Ventura Ultra.

Recently, we also received a sample of another SSD controller-based flash drive. Putting that drive through our throttling test revealed even worse results compared to the Mushkin Ventura Ultra. Upon our feedback, the manufacturer is contemplating a firmware update. In this piece, we will refer to the unit as ‘BrandX 240GB’ (a complete review of the unit with benchmark numbers will be published once the manufacturer provides us a resolution for the throttling issue). The list of DAS units put under the performance consistency scanner is provided below.

  • Mushkin Ventura Ultra 120GB
  • BrandX 240GB
  • Corsair Voyager GTX 256GB
  • LaCie Rugged Thunderbolt 500GB
  • Mushkin Atom 64GB

Performance Consistency Evaluation Methodology

Our testing methodology for DAS units takes into consideration the usual use-case for such devices. The most common usage scenario is transfer of large amounts of photos and videos to and from the unit. In order to tackle this, we created three test folders with the following characteristics:

  • Photos: 15.6 GB collection of 4320 photos (RAW as well as JPEGs) in 61 sub-folders
  • Videos: 16.1 GB collection of 244 videos (MP4 as well as MOVs) in 6 sub-folders
  • BR: 10.7 GB Blu-ray folder structure of the IDT Benchmark Blu-ray (the same that we use in our robocopy tests for NAS systems)

These folders were put in a RAM drive on our testbed. Using robocopy, each folder was copied from the RAM drive to the DAS (write benchmark) and back to the RAM drive (read benchmark). This process was repeated thrice. To check for performance consistency, we instrumented our benchmark script to record the flash drive’s read and write transfer rates while the robocopy process took place in the background. For supported drives, we also recorded the internal temperature of the drive during the process.

Evaluation Results

Aspects that may influence performance consistency include thermal throttling and firmware caps on access rates to avoid overheating or other similar scenarios. This aspect is an important one, as the last thing that users want to see when copying over, say, 100 GB of data to the flash drive, is the transfer rate going to USB 2.0 speeds. The graphs below show the speeds observed during our real-world DAS suite processing with the instrumented benchmark script. The first three sets of writes and reads correspond to the photos suite. A small gap (for the transfer of the videos suite from the primary drive to the RAM drive) is followed by three sets for the next data set. Another small RAM-drive transfer gap is followed by three sets for the Blu-ray folder.

An important point to note here is that each of the first three blue and green areas correspond to 15.6 GB of writes and reads respectively. Throttling, if any, is apparent within the processing of the photos suite itself.

Mushkin Ventura Ultra 120GBBrandX 240GBCorsair Voyager GTX 256GBLaCie Rugged Thunderbolt 500GBMushkin Atom 64GB

The Mushkin Ventura Ultra 120GB does throttle after the S.M.A.R.T temperature report reaches 45 C. In our testbed (ambient at 22C), this temperature was reached after approximately 18 GB of writes and 16 GB of reads without any interruptions. Beyond this, the write speeds got capped around 80 MBps (compared to the non-throttled 100 MBps), while the reads got capped around 100 MBps (compared to the non-throttled 140 MBps). The throttling allowed the temperature to stay below 50 C throughout our benchmark run.

BrandX, though, behaved differently. The temperature in the steady state settled around 53 C. Unfortunately, the throttling (after a similar amount of data transfer as the Mushkin Ventura Ultra) resulted in the write speeds getting capped at a paltry 20 MBps. The reads managed to remain around 80 MBps.

The Mushkin Atom is not meant to take the performance crown, and the advertised write speeds are in the 20 MBps range. Our benchmark run ran at a steady 15 MBps, while the reads didn’t get throttled at all in our evaluation process.

The LaCie Rugged Thunderbolt drive is not a flash drive (unlike the other units considered in this piece). It is a bona-fide SSD in a USB 3.0 / Thunderbolt hybrid enclosure. It is no wonder that the transfer rates were hardly affected by the relatively small amount of data traffic. The real winner of our evaluation was the Corsair Voyager GTX. There is no performance degradation even after more than 120 GB of continuous traffic. It must also be remembered that we ran our robocopy benchmark with multiple queue depths in our original review (that was more than 480 GB of continuous traffic) and the table showed remarkable consistency. The instrumented benchmark run confirmed the same.

Concluding Remarks

Coming to the business end of the piece, it must be remembered that the capabilities must be compared keeping the price in mind. If cost is not an issue and the form factor is not an issue, the LaCie Rugged Thunderbolt is the hands down winner for a fast portable DAS unit with consistent performance. If cost is the primary concern, and one doesn’t mind a little bit of throttling after around 18 GB of continuous write traffic, the Mushkin Ventura Ultra is a great choice at less than $0.6/GB. The middle ground (right balance of cost and fast, consistent performance, along with a traditional flash drive form factor) is held by the Corsair Voyager GTX. At more than $0.85/GB, it also serves as a reminder that one gets what they pay for.

Intel Launches M.2 and Larger Capacity 2.5" SSD DC S3500 Models

Intel Launches M.2 and Larger Capacity 2.5″ SSD DC S3500 Models

Two years ago Intel released the DC S3500 and S3700 SSDs that marked as the company’s first truly in-house SATA 6Gbps SSDs. Today Intel is adding M.2 form factor to the S3500 lineup along with higher capacity options in the regular 2.5″ size.

The original S3500 came in capacities ranging from as small as 80GB to up to 800GB, but with the increasing demand for high density storage, Intel is adding 1.2TB and 1.6TB offerings. There is no change in the Flash Translation Layer (FTL) design to accommadate the extra NAND nor are there any other firmware or hardware changes, so the new models simply have more NAND on the board.

The 1.2TB and 1.6TB models provide higher randon write speeds, which makes sense given the additional NAND. Typically steady-state random write performance increases with more NAND thanks to increased parallelism, although only to a certain threshold since more NAND also adds tracking overhead. In the case of the S3500, the 1.2TB model appears to be the sweetspot by offering 18.5K random write IOPS, whereas the 1.6TB SKU drops that figure to 14.6K IOPS. Moreover, write endurance scales nearly linearly with the capacity as the 1.2TB is rated at 660TB and the 1.6TB at 880TB, while the existing 600GB and 800GB models are rated at 330TB and 450TB respectively. 

The M.2 version of the S3500 is one of the first enterprise-class SSDs in M.2 form factor. Feature wise the M.2 models are equal to their 2.5″ counterparts, meaning that there is full power-loss protection, end-to-end data protection and AES-256 encryption. The M.2 models come in M.2 2280 form factor and due to the space constraints, only 80GB, 120GB and 340GB capacities are available. 

The M.2 models actually carry slightly higher random write speeds compared to their 2.5″ counterparts, although that comes at the slight expense of random read performance. Since there is no metal chassis to help with heat dissipation, the M.2 models are a bit more exposed to thermal throttling, although Intel assured that as long as there is enough airflow in the system the drive should stay within its operating temperatures and not throttle itself.

At this stage M.2 hasn’t really been adopted by the enterprise space yet, but there is a growing interest for it due to the space savings. Intel told me that they are in talks with multiple OEMs about integrating the M.2 S3500 to servers, but as of now the main market for the S3500 is embedded devices. This includes devices like ATMs, slot machines and cash registers, which are transitioning to SSDs to save space and increase reliability. The S3500 is certainly a good fit for that segment with its enterprise-class feature set, whereas most M.2 drives on the market lack features like full power-loss protection. 

While many consumer drives are selling at below $0.50/GB, the enterprise features and additional validation increase the cost of the S3500. At right below the $1 per GB mark, the S3500 is, however, a rather competitive drive in the entry-level enterprise segment.

Intel Launches M.2 and Larger Capacity 2.5" SSD DC S3500 Models

Intel Launches M.2 and Larger Capacity 2.5″ SSD DC S3500 Models

Two years ago Intel released the DC S3500 and S3700 SSDs that marked as the company’s first truly in-house SATA 6Gbps SSDs. Today Intel is adding M.2 form factor to the S3500 lineup along with higher capacity options in the regular 2.5″ size.

The original S3500 came in capacities ranging from as small as 80GB to up to 800GB, but with the increasing demand for high density storage, Intel is adding 1.2TB and 1.6TB offerings. There is no change in the Flash Translation Layer (FTL) design to accommadate the extra NAND nor are there any other firmware or hardware changes, so the new models simply have more NAND on the board.

The 1.2TB and 1.6TB models provide higher randon write speeds, which makes sense given the additional NAND. Typically steady-state random write performance increases with more NAND thanks to increased parallelism, although only to a certain threshold since more NAND also adds tracking overhead. In the case of the S3500, the 1.2TB model appears to be the sweetspot by offering 18.5K random write IOPS, whereas the 1.6TB SKU drops that figure to 14.6K IOPS. Moreover, write endurance scales nearly linearly with the capacity as the 1.2TB is rated at 660TB and the 1.6TB at 880TB, while the existing 600GB and 800GB models are rated at 330TB and 450TB respectively. 

The M.2 version of the S3500 is one of the first enterprise-class SSDs in M.2 form factor. Feature wise the M.2 models are equal to their 2.5″ counterparts, meaning that there is full power-loss protection, end-to-end data protection and AES-256 encryption. The M.2 models come in M.2 2280 form factor and due to the space constraints, only 80GB, 120GB and 340GB capacities are available. 

The M.2 models actually carry slightly higher random write speeds compared to their 2.5″ counterparts, although that comes at the slight expense of random read performance. Since there is no metal chassis to help with heat dissipation, the M.2 models are a bit more exposed to thermal throttling, although Intel assured that as long as there is enough airflow in the system the drive should stay within its operating temperatures and not throttle itself.

At this stage M.2 hasn’t really been adopted by the enterprise space yet, but there is a growing interest for it due to the space savings. Intel told me that they are in talks with multiple OEMs about integrating the M.2 S3500 to servers, but as of now the main market for the S3500 is embedded devices. This includes devices like ATMs, slot machines and cash registers, which are transitioning to SSDs to save space and increase reliability. The S3500 is certainly a good fit for that segment with its enterprise-class feature set, whereas most M.2 drives on the market lack features like full power-loss protection. 

While many consumer drives are selling at below $0.50/GB, the enterprise features and additional validation increase the cost of the S3500. At right below the $1 per GB mark, the S3500 is, however, a rather competitive drive in the entry-level enterprise segment.

Mushkin Ventura Ultra USB 3.0 120GB Flash Drive Capsule Review

Mushkin Ventura Ultra USB 3.0 120GB Flash Drive Capsule Review

The rise of USB 3.0 as a high speed interface for PCs and the increasing affordability of flash memory has led to some very interesting products. USB flash drives are a dime a dozen, but there is scope for manufacturers to differentiate themselves. We have already looked at the Corsair Flash Voyager GTX, which brought in a Phison controller into the flash driver market. Despite the presence of a USB 3.0 – SATA bridge chip, the unit supported all the important SSD functionalities such as TRIM and overprovisioning. Prior to the Corsair’s Voyager GTX introduction, Mushkin actually demonstrated and brought a SSD controller-based USB 3.0 flash drive to the market in the Ventura Ultra series.

The Mushkin Ventura Ultra series is based on the Sandforce SF-2281 controller. SSDs based on that controller have been in the market for quite some time. The firmware is quite mature now. The bridge chip used in the drives is the ASMedia 1053E. Mushkin also tagged along a 64 GB Atom USB 3.0 drive (that we briefly saw at CES earlier this year) with the Ventura Ultra review sample. The rest of the review will present the DAS (direct-attached storage) benchmark numbers from our evaluation of the two devices, along with a few observations.

Testbed Setup and Testing Methodology

Evaluation of DAS units on Windows is done with the testbed outlined in the table below. For devices with USB 3.0 connections (such as the Ventura Ultra and Atom flash drives being considered today), we utilize the USB 3.0 port directly hanging off the PCH.

AnandTech DAS Testbed Configuration
Motherboard Asus Z97-PRO Wi-Fi ac ATX
CPU Intel Core i7-4790
Memory Corsair Vengeance Pro CMY32GX3M4A2133C11
32 GB (4x 8GB)
DDR3-2133 @ 11-11-11-27
OS Drive Seagate 600 Pro 400 GB
Optical Drive Asus BW-16D1HT 16x Blu-ray Write (w/ M-Disc Support)
Add-on Card Asus Thunderbolt EX II
Chassis Corsair Air 540
PSU Corsair AX760i 760 W
OS Windows 8.1 Pro
Thanks to Asus and Corsair for the build components

The full details of the reasoning behind choosing the above build components can be found here. The list of DAS units used for comparison purposes is provided below:

  • Mushkin Ventura Ultra 120GB USB 3.0
  • Mushkin Atom 64GB USB 3.0
  • Corsair Voyager GTX 256GB USB 3.0
  • LaCie Rugged Thunderbolt 500GB (SSD version, over USB 3.0)

Synthetic Benchmarks – ATTO and CrystalDiskMark

Mushkin claims read and write speeds of 445 MBps and 440 MBps respectively for the 120GB Ventura Ultra. These are backed up by the ATTO benchmarks provided below. The Atom version claims read and write speeds of 120 and 20 MBps. We don’t see anywhere close to those numbers in ATTO, though some of our other benchmarks (detailed further down) do come a bit close. Unfortunately, the numbers provided by ATTO are not directly applicable to most real-life scenarios. It must be noted that the SF-2281 drives tend to perform well with compressible data (as the controller does on-the-fly compression and de-duplication to reduce NAND wear). The data used by ATTO for testing can’t be configured and might turn out to be compressible, tending to favor the SandForce SF-2281 controller more.

Mushkin Ventura Ultra 120GBCorsair Voyager GTX 256GBLaCie Rugged Thunderbolt 500GBMushkin Atom 64GB

CrystalDiskMark, despite being a canned benchmark, provides a better estimate of the performance range with a selected set of numbers. As evident from the screenshot below, the performance can dip to as low as 14 MBps for the Ventura Ultra for 4K random accesses. The less said about the Atom drive here, the better it is. In any case, the Atom is not being marketed for its transfer rate capabilities.

Mushkin Ventura Ultra 120GBCorsair Voyager GTX 256GBLaCie Rugged Thunderbolt 500GBMushkin Atom 64GB

robocopy and PCMark 8 Storage Bench

Our testing methodology for DAS units also takes into consideration the usual use-case for such devices. The most common usage scenario is transfer of large amounts of photos and videos to and from the unit. The minor usage scenario is importing files directly off the DAS into a multimedia editing program such as Adobe Photoshop.

In order to tackle the first use-case, we created three test folders with the following characteristics:

  • Photos: 15.6 GB collection of 4320 photos (RAW as well as JPEGs) in 61 sub-folders
  • Videos: 16.1 GB collection of 244 videos (MP4 as well as MOVs) in 6 sub-folders
  • BR: 10.7 GB Blu-ray folder structure of the IDT Benchmark Blu-ray (the same that we use in our robocopy tests for NAS systems)

robocopy - Photos Read

robocopy - Photos Write

robocopy - Videos Read

robocopy - Videos Write

robocopy - Blu-ray Folder Read

robocopy - Blu-ray Folder Write

The 500 GB LaCie Rugged Thunderbolt wins all benchmark hands down. However, as we shall see further down, it is also the costliest of the lot. The form factor also doesn’t match up to those of a typical flash drive. Between the Ventura Ultra and the Voyager GTX, there isn’t much separation in the write benchmarks. However, the reads present a different story – the Ventura Ultra lags by quite a bit. We also see the 64GB Atom provide 100 MBps+ read speeds for some workloads.

For the second use-case, we take advantage of PC Mark 8’s storage bench. The storage workload involves games as well as multimedia editing applications. The command line version allows us to cherry-pick storage traces to run on a target drive. We chose the following traces.

  • Adobe Photoshop (Light)
  • Adobe Photoshop (Heavy)
  • Adobe After Effects
  • Adobe Illustrator

Usually, PC Mark 8 reports time to complete the trace, but the detailed log report has the read and write bandwidth figures which we present in our performance graphs. Note that the bandwidth number reported in the results don’t involve idle time compression. Results might appear low, but that is part of the workload characteristic. Note that the same testbed is being used for all DAS units. Therefore, comparing the numbers for each trace should be possible across different DAS units.

robocopy - Photoshop Light Read

robocopy - Photoshop Light Write

robocopy - Photoshop Heavy Read

robocopy - Photoshop Heavy Write

robocopy - After Effects Read

robocopy - After Effects Write

robocopy - Illustrator Read

robocopy - Illustrator Write

In these benchmarks, the Ventura Ultra and Voyager GTX are more evenly matched, trading honors depending on the workload. It must be remembered that the toss-up between the Ventura Ultra and the Voyager GTX is not exactly an apples-to-apples comparison because of the difference in the capacities.

Concluding Remarks

Coming to the business end of the review, the Ventura Ultra keeps up Mushkin’s claims of being one of the fastest flash drives in the market. The icing on the cake is the presence of a real SSD controller in the form of the SandForce SF2281 inside.

The performance of the drive leaves us with no doubt that it would be a great portable OS drive, even though Mushkin doesn’t advertise it for that purpose. The Atom series, on the other hand, is nothing great when it comes to performance. However, the miniature size is the main selling point here.

Minor points of concern include Mushkin’s refusal to divulge the flash memory configuration being used (as these could potentially change in future production runs) and the inability to recognize and take actions on the drive based on S.M.A.R.T features such as TRIM. That said, it is not really clear whether TRIM on SF2281 drives is really worth the trouble. In any case, there is inherent over-provisioning in the drive and the typical flash drive use-case rarely fills up the drive enough to cause extensive performance degradation.

Price per GB

The pricing aspect is the final item on the table. The LaCie Rugged Thunderbolt is obviously costly because of the availability of dual interfaces (both USB 3.0 and Thunderbolt). The Atom comes in as the cheapest of the lot (after all, it is a typical flash drive). The Ventura Ultra, on a per-GB basis, is definitely more attractive than the Corsair Voyager GTX. However, long-term performance consistency for heavy use-cases (not the typical flash drive usage scenario) may be in favor of the Corsair unit (thanks to TRIM support). Over-provisioning can be configured for the Voyager GTX also.

In terms of absolute pricing, the Mushkin Ventura Ultra is available for $70. The 64GB Atom drive retails for $29. A typical SF-2281-based 120GB SSD in a 2.5″ bus-powered USB 3.0 enclosure retails for around $90. The form factor of the Ventura Ultra, fortunately, doesn’t carry a premium. This makes it easy to recommend it to users looking for a flash drive that can take full advantage of USB 3.0 speeds.