It's been a year and a half since OCZ went bankrupt and Toshiba acquired its assets. In this time we have seen OCZ transition all of its products to Toshiba NAND and we have also discussed how Toshiba has helped OCZ with its quality/validation processes, but quite frankly we haven't seen anything truly concrete coming out of the partnership. The Trion 100 is here to change that as it's the first Toshiba built drive that will retail under OCZ's brand.

The Trion 100 is a significant signal of the companies' strategy moving forward. While Toshiba is letting OCZ run its business independently, the two work closely together on new products and roadmaps to maximize resources and avoid overlaps. For TLC the companies decided to utilize Toshiba's IP because it had been developing a TLC SSD for years and was simply much longer in the process compared to OCZ. The Barefoot 3 doesn't have proper TLC support, so OCZ would have needed to either design a new controller or use a third party one, which isn’t as economical as using an in-house design. 

OCZ wasn't involved in the development of the Trion 100, but it did help Toshiba to validate the drive. Basically, Toshiba approached OCZ in Design Verification Test (see our fab tour article for details) phase when it already had working samples and OCZ provided Toshiba with general feedback and suggestions for bug fixes through its own internal testing. None of OCZ's firmware engineers were involved in the Trion 100 project, though, as the design and development were executed solely by Toshiba. Manufacturing and run-in testing is also handled by Toshiba, so in all honesty the Trion 100 is a Toshiba drive that retails under OCZ as Toshiba doesn’t directly engage in the retail market.

Since Toshiba only does OEM sales, it isn't used to discussing its technology in detail, especially with media and the public. In the OEM market it's fine to sell a "black box" and only disclose the necessary specs (like performance and endurance) in a data sheet because ultimately the OEM customers are buying a solution with the underlying technology being a secondary interest (I would argue that most PC OEMs don't even fully understand the technology). 

960GB Trion 100 on the left, 960GB Corsair Neutron XT on the right

As a result, the technical details of the drive are very scarce. The TC58 controller is supposedly made by Toshiba and so is the firmware, but Toshiba isn't willing to disclose any specifics of the architecture. However, I discovered that the PCB layout closely resembles the Phison S10, and the relation is substantial enough that I doubt it's a coincidence. There are only minor differences in resistor and capacitor placements, but when focusing on the area under the controller the two are undoubtedly a match. The timing is another clue because at Computex Phison told me that the company is a week or two away from finalizing the TLC firmware for the S10 and now a few weeks later Toshiba puts out its first TLC drive with nearly identical look to Phison S10 drives.

Officially Toshiba and OCZ are saying it’s a Toshiba controller, although OCZ is aware of the resemblance, but obviously the company can’t differ from Toshiba’s word. I did let OCZ know about my concerns regarding another rebrand (the Indilinx Everest controller turned out to be a Marvell silicon after all) and what it may do to the recovering OCZ brand, but OCZ’s hands are tied because the Trion 100 is fundamentally a Toshiba drive. OCZ has always been quite open about its technologies (Barefoot 3 for example), but as Toshiba has been doing business in a certain way for many years it will take time for the two to adapt. In the end, the Trion 100 has significant educational value to the Toshiba-OCZ organization because this is the first time that a Toshiba drive is specifically marketed for the retail market. 

Toshiba and Phison have a very close relationship, so a co-developed controller is hardly a surprise, especially since Toshiba has done it before with Marvell and JMicron. Toshiba is one of the few companies that have (limited) access to Phison's firmware source codes and all Phison drives are manufactured in Toshiba's factories (Phison doesn't sell drives under its own brand, but it provides fully assembled drives to companies such as Corsair and Mushkin).  

To me the biggest question is how customized the firmware is and how much it differs from the stock S10 firmware that Phison will be shipping to its own OEMs. I think it's safe to say that the silicon itself is the same because a separate die would be costly, but I suspect Toshiba has had a significant impact on the firmware. I don’t have any Phison S10 with TLC at hand at the moment, so we’ll have to wait a little longer before we can investigate how big of a difference Toshiba’s firmware makes.

OCZ Trion 100 Specifications
Capacity	120GB	240GB	480GB	960GB
Controller	Toshiba TC58
NAND	Toshiba A19nm 128Gbit TLC
Sequential Read	550MB/s	550MB/s	550MB/s	550MB/s
Sequential Write	450MB/s	520MB/s	530MB/s	530MB/s
4KB Random Read	79K IOPS	90K IOPS	90K IOPS	90K IOPS
4KB Random Write	25K IOPS	43K IOPS	54K IOPS	64K IOPS
Endurance	30TB	60TB	120TB	240TB
DevSleep Power	6mW
Idle Power	830mW
Max Power	4.8W
Encryption	N/A
Warranty	Three years
MSRP	$57	$88	$185	$370

The TC58 doesn’t support LDPC error correction (and neither does the S10 for that matter), but utilizes Toshiba’s QSBC (Quadruple Swing-By Codes) ECC. Since Toshiba is tight lipped about any of its in-house technologies, the details surrounding QSBC are scarce, but I would speculate that it's something in-between BCH and LDPC error corrections. While LDPC offers higher error correction capability, it’s also slower as soft decoding is a time consuming process, so QSBC should provide better performance while still providing enough error correction capability to enable the use of TLC NAND.

OCZ Trion 100 NAND Configurations
Capacity	120GB	240GB	480GB	960GB
SLC Cache Size	1.8GB	3.6GB	7.2GB	14.4GB
Raw NAND Capacity	128GiB	256GiB	512GiB	1,024GiB
# of NAND Packages	4	4	4	4
# of Die per Package	2x16GiB	4x16GiB	8x16GiB	16x16GiB

The Trion 100 uses pseudo-SLC cache to for improving write performance. The NAND is a specific SKU with pseudo-SLC capability (instead of simply writing to the lower pages), but unlike what SanDisk is doing in the Ultra II, Toshiba’s NAND doesn’t have a fixed SLC portion with different transistor characteristics. The size of the SLC cache is fixed, though, and it’s 1.5% of the total NAND capacity. That’s similar to what we have seen in other TLC drives and since client workloads are bursty by nature there isn’t really a need for more than a few gigabytes of SLC. There are some write optimizations done to reduce write amplification when writing from SLC to TLC, but again even OCZ isn’t aware of the specifics.

Despite the use of TLC NAND, the Trion 100 boasts excellent endurance. A part of that has to do with different validation workload because Toshiba uses JEDEC’s client workload to specify endurance, whereas OCZ provides worst-case figures based on a 4KB random write workload with substantially higher write amplification. This stems from the enterprise space where OCZ has been using 4KB random writes ever since the company started reporting endurance and in order to align with the enterprise numbers OCZ has decided to use the same workload for client drives as well, despite the fact that it’s arguably an overkill. OCZ is currently validating the rest of its drives with the JEDEC client spec and in the future we will likely see both figures, which I find to be a good strategy because it also gives us an insight to the write amplification. 

The Trion uses Toshiba's A19nm 128Gbit TLC NAND, which is manufactured using the same underlying technology as the NAND used in SanDisk's Ultra II since the two companies have a NAND joint-venture. The most interesting aspect is that the 960GB model uses 16-die packages, which we haven't seen from Toshiba in a shipping product yet. Toshiba's NAND is actually packaged by a company called PTI, which is the same company that assembles SSDs for OCZ and a variety of other vendors.

AnandTech 2015 SSD Test System
CPU	Intel Core i7-4770K running at 3.5GHz (Turbo & EIST enabled, C-states disabled)
Motherboard	ASUS Z97 Deluxe (BIOS 2205)
Chipset	Intel Z97
Chipset Drivers	Intel 10.0.24+ Intel RST 13.2.4.1000
Memory	Corsair Vengeance DDR3-1866 2x8GB (9-10-9-27 2T)
Graphics	Intel HD Graphics 4600
Graphics Drivers	15.33.8.64.3345
Desktop Resolution	1920 x 1080
OS	Windows 8.1 x64

• Thanks to Intel for the Core i7-4770K CPU
• Thanks to ASUS for the Z97 Deluxe motherboard
• Thanks to Corsair for the Vengeance 16GB DDR3-1866 DRAM kit, RM750 power supply, Hydro H60 CPU cooler and Carbide 330R case

Performance Consistency

We've been looking at performance consistency since the Intel SSD DC S3700 review in late 2012 and it has become one of the cornerstones of our SSD reviews. Back in the days many SSD vendors were only focusing on high peak performance, which unfortunately came at the cost of sustained performance. In other words, the drives would push high IOPS in certain synthetic scenarios to provide nice marketing numbers, but as soon as you pushed the drive for more than a few minutes you could easily run into hiccups caused by poor performance consistency.

Once we started exploring IO consistency, nearly all SSD manufacturers made a move to improve consistency and for the 2015 suite, I haven't made any significant changes to the methodology we use to test IO consistency. The biggest change is the move from VDBench to Iometer 1.1.0 as the benchmarking software and I've also extended the test from 2000 seconds to a full hour to ensure that all drives hit steady-state during the test.

For better readability, I now provide bar graphs with the first one being an average IOPS of the last 400 seconds and the second graph displaying the standard deviation during the same period. Average IOPS provides a quick look into overall performance, but it can easily hide bad consistency, so looking at standard deviation is necessary for a complete look into consistency.

I'm still providing the same scatter graphs too, of course. However, I decided to dump the logarithmic graphs and go linear-only since logarithmic graphs aren't as accurate and can be hard to interpret for those who aren't familiar with them. I provide two graphs: one that includes the whole duration of the test and another that focuses on the last 400 seconds of the test to get a better scope into steady-state performance.

TLC SSDs in general don't perform well under sustained random workloads and the Trion 100 is no exception. It seems that with TLC SSDs we have to accept a drop from 5K to 2-3K IOPS, which is still fine for basic client workloads since most SSDs from 3-4 years ago couldn't even match that.

The consistency isn't great either, but it's not particularly poor either when compared agains Silicon Motion drives.

Default	OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBSMI 2256 Reference Design 500GBCorsair Neutron XT 480GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GB
25% Over-Provisioning	OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBSMI 2256 Reference Design 500GBCorsair Neutron XT 480GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GB

Looking at the performance over time, we can see that the baseline performance hovers at about 1,000 IOPS with frequent peaks occuring at 5K to 10K IOPS. That's actually very similar to the Neutron XT's (Phison S10) graph because the baseline is also 1,000 IOPS, although the peaks are higher and more frequent, but it further reassures that the underlying firmware architecture is similar. Increasing over-provisioning doesn't increase the baseline performance, but it does make peak performance moments more frequent and higher (from 10K IOPS to 25K IOPS). 

Default	OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBSMI 2256 Reference Design 500GBCorsair Neutron XT 480GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GB
25% Over-Provisioning	OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBSMI 2256 Reference Design 500GBCorsair Neutron XT 480GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GB

AnandTech Storage Bench - The Destroyer

The Destroyer has been an essential part of our SSD test suite for nearly two years now. It was crafted to provide a benchmark for very IO intensive workloads, which is where you most often notice the difference between drives. It's not necessarily the most relevant test to an average user, but for anyone with a heavier IO workload The Destroyer should do a good job at characterizing performance. For full details of this test, please refer to this article.

The Trion is evidently not designed for intensive IO workloads like our The Destroyer and that's clear in the results. It's quite a bit slower than any of the other drives we have tested, including the TLC based Silicon Motion SM2256.

Fortunately the share of high latency IOs is tolerable and despite the high average latency the Trion at least doesn't completely stop processing host IOs. 

It looks like the Trion is doing a lot of background garbage collection because despite the low performance, the power consumption is very high. For desktop users that's a non-issue, but for mobile the Trion may not be the best pick.

AnandTech Storage Bench - Heavy

While The Destroyer focuses on sustained and worst-case performance by hammering the drive with nearly 1TB worth of writes, the Heavy trace provides a more typical enthusiast and power user workload. By writing less to the drive, the Heavy trace doesn't drive the SSD into steady-state and thus the trace gives us a good idea of peak performance combined with some basic garbage collection routines. For full details of the test, please refer to the this article.

Ouch, this doesn't look too good. Even out of the TLC drives the Trion 100 is the slowest of the bunch and the difference isn't marginal either (~25% drop in data rate at 240GB compared to the Ultra II). 

The latencies actually show a worse phenomenon as MLC based drives (even the Neutron XT) all easily stay below 800µs, whereas TLC drives start at 1,000µs and the Trion even goes above 1,500µs. 

Unsurprisingly, the high average latency shows up as an increased share of high latency IOs.

Not only is the Trion slower than the rest, it also consumes more power. To be honest, I don't mind a lower performance drive if the decreased performance translates to lower power consumption, but if that doesn't happen then the whole design is simply inefficient like in Trion's case. As I've mentioned before, TLC is inherently less power efficient so increased power consumption over MLC SSDs is expected, but I still think the Trion could use some better optimization. 

AnandTech Storage Bench - Light

The Light trace is designed to be an accurate illustration of basic usage. It's basically a subset of the Heavy trace, but we've left out some workloads to reduce the writes and make it more read intensive in general. Please refer to this article for full details of the test.

Under our Light workload the Trion fortunately performs better. Ultimately the average Joes are the target market for the Trion, so it's good to see it being at least relatively competitive especially with the SMI2256, although it is still outperformed by the 850 EVO and Ultra II. 

The power consumption is also much more reasonable now.

Random Read Performance

For full details of how we conduct our Iometer tests, please refer to this article.

Random read speeds already shed some light to why the Trion didn't perform so well in our Storage Benches as it's simply way behind any other drive. 

While the Trion isn't the most power hungry drive, it certainly consumes more than it should given its low performance.

OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBOCZ ARC 100 240GBCorsair Neutron XT 480GBCrucial BX100 120GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO M.2 120GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GBSMI 2256 Reference Design 500GB

The Trion starts from the bottom because its QD1 performance at 16-20MB/s is significantly below the rest and the scaling isn't aggressive enough to make up for the difference. High queue depth performance isn't outstanding either, but luckily that has much less relevance in the real world.

Random Write Performance

Unfortunately random write performance doesn't show any better signs. The Trion is again at the bottom of the chart by a quite substantial margin.

Power consumption, on the other hand, is fairly high, making the Trion one of the least efficient drives.

OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBOCZ ARC 100 240GBCorsair Neutron XT 480GBCrucial BX100 120GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO M.2 120GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GBSMI 2256 Reference Design 500GB

There is effectively no scaling with queue depth at all, which I find strange. I suspect this is a limitation of the firmware because with 1TB of TLC NAND there should be enough NAND bandwidth to scale, but obviously if the firmware is inefficiently designed that doesn't happen.

Sequential Read Performance

For full details of how we conduct our Iometer tests, please refer to this article.

Fortunately sequential read performance is better than random, although for some reason the performance decreases as the capacity goes up. It seems like the controller may have trouble keeping up with the increasing number of LBAs to track, or frankly it could be just poor optimization too because the Neutron XT has no similar performance issues despite the similar underlying architecture. Another explanation could be increased latency from higher density NAND packages as there is some performance penalty due to longer wiring, although Samsung doesn't have any performance issues with its high density packages.

Unlike the performance, power consumption does go up with capacity, making the higher capacities less efficient.

OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBOCZ ARC 100 240GBCorsair Neutron XT 480GBCrucial BX100 120GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO M.2 120GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GBSMI 2256 Reference Design 500GB

The reason for poor performance lies in bad scaling because it takes a queue depth of 16 until the 480GB and 960GB models reach their peak performance.

Sequential Write Performance

Sequential write performance, on the other hand, is as poor as random write performance. I'm again surprised that the higher capacities present only marginal increase, whereas in write performance the additional NAND should help to distribute writes to a larger number of die for increased performance.

Power is again pretty high, though, especially when considering the performance against other drives.

OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBOCZ ARC 100 240GBCorsair Neutron XT 480GBCrucial BX100 120GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO M.2 120GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GBSMI 2256 Reference Design 500GB

Mixed Random Read/Write Performance

For full details of how we conduct our Iometer tests, please refer to this article.

In mixed random performance the Trion at least isn't the only drive that performs poorly because the Ultra II is nearly as bad, but that's is a very little delight.

Power follows the same pattern as we saw in the two previous pages: it's not specifically higher than others but when performance is taken into account the efficiency is rather poor.

OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBOCZ ARC 100 240GBCorsair Neutron XT 480GBCrucial BX100 120GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO M.2 120GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GBSMI 2256 Reference Design 500GB

In terms of scaling the Trion is just generally slow; there's no particular distribution that stands out except pure writes but that was expected given the poor random write performance in the first place.

Mixed Sequential Read/Write Performance

In mixed sequential performance, once again, the Trion is at the bottom and this time the power consumption is also higher than the rest.

OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBOCZ ARC 100 240GBCorsair Neutron XT 480GBCrucial BX100 120GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ ARC 100 240GBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO M.2 120GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GBSMI 2256 Reference Design 500GB

ATTO - Transfer Size vs Performance

ATTO is a handy tool for quickly measuring performance across various transfer sizes and it's also freeware that can easily be run by the end-user.

OCZ Trion 100 240GBOCZ Trion 100 480GBOCZ Trion 100 960GBCorsair Neutron XT 480GBCrucial BX100 120GBCrucial BX100 250GBCrucial BX100 500GBCrucial BX100 1TBCrucial MX100 512GBCrucial MX200 250GBCrucial MX200 500GBCrucial MX200 1TBOCZ Vector 180 240GBOCZ Vector 180 480GBOCZ Vector 180 960GBSamsung 850 EVO M.2 120GBSamsung 850 EVO mSATA 250GBSamsung 850 EVO M.2 500GBSamsung 850 EVO mSATA 1TBSamsung 850 EVO 1TBSamsung SSD 850 Pro 512GBSanDisk Extreme Pro 480GBSanDisk Ultra II 240GBSMI 2256 Reference Design 500GB

The 240 GB drive seems to have an issue when it comes to writes here, showing variable performance across any transfer size due to the SLC buffer being filled and emptied during the process.

AS-SSD Incompressible Sequential Performance

Similar to ATTO, AS-SSD is freeware as well and uses incompressible data for all of its transfers, making it a valuable tool when testing drives with built-in compression engines (e.g. SandForce).

When an SSD vendor quotes peak read and write speeds, these are the numbers typically referred to, despite its supposed relevance to regular workloads.

Idle Power Consumption

Since we truncate idle times to 25µs in our Storage Bench traces, they don't give a fully accurate picture of real world power consumption as idle power consumption is not taken properly into account. Hence I'm still reporting idle power consumption as a separate benchmark because it's one of the most critical metrics when it comes evaluating an SSD for mobile use.

Unfortunately I still don't have a way to test DevSleep power consumption due to lack of platform support, but my testbed supports HIPM+DIPM power commands (also referred to as Slumber power), so the results give a rather accurate picture of real-world idle power consumption. 

The Trion is the first OCZ SSD to include support for low-power states and that's clearly visible in our results. It's not the lowest power drive we have tested, but falls in the same range as the majority of drives. 

TRIM Validation

The move from Windows 7 to 8.1 introduced some problems with the methodology we have previously used to test TRIM functionality, so I had to come up with a new way to test. I tested a couple of different methods, but ultimately I decided to go with the easiest one that can actually be used by anyone. The software is simply called trimcheck and it was made by a developer that goes by the name CyberShadow in GitHub. 

Trimcheck tests TRIM by creating a small, unique file and then deleting it. Next the program will check whether the data is still accessible by reading the raw LBA locations. If the data that is returned by the drive is all zeros, it has received the TRIM command and TRIM is functional. 

Final Words

The bar for TLC SSDs was set by Samsung in 2012 and frankly that bar was set very high. As a company that develops everything in-house, Samsung has a massive advantage when it comes to new technologies because its controller and NAND teams can collaborate in a way that's not possible for other companies. In the end, TLC is both a NAND and a controller game because you need to do tricks on both sides to design a drive that can provide close to MLC-like performance, while still keeping the cost benefits of TLC. 

To date, nobody has been able to cross the bar Samsung set and neither does the Trion 100. In fact, the Trion is the slowest modern TLC drive we've tested and in some cases it is so by a quite hefty margin. Typical to TLC drives, the Trion falls short under anything IO intensive. It does perform okay (although it's still the slowest) in our Light trace, but as soon as the IO workload is increased the Trion begins to hiccup. Since the Trion is an entry-level drive, that's not a matter of life and death because in most cases it will only be subjected to basic client workloads, which are far from being IO intensive, perhaps negating the issue and moving the contest more to price. 

The specific areas that need improvement are low queue depth read performance and write performance in general. Random read performance especially at queue depth of 1 is lacking and up to 50% lower compared to competing drives, and for some reason the sequential read performance has queue depth scaling issues at 480GB and 960GB where the drive needs a very high queue depth to reach its maximum throughput. Despite using a pseudo-SLC cache for writes, the write performance in every scenario is underwhelming and one of the key issues is the fact that there is absolutely no performance scaling with queue depth regardless of the capacity. 

The poor performance also translates to power inefficiency. Because the drive spends more time processing each IO, it will end up idling less than a faster drive would and as we saw in the trace-based tests that results in higher total power consumed than any other drive we have tested. For desktop users this is hardly an issue, but I would suggest mobile users to look elsewhere, namely BX100 and 850 EVO.

Amazon Price Comparison (7/9/2015)
	120/128GB	240/250/256GB	480/500/512GB	960GB/1TB
OCZ Trion 100	$60	$90	$180	$360
Crucial BX100	$65	$90	$180	$380
OCZ ARC 100	$54	$89	$170	-
Samsung 850 EVO	$68	$98	$162	$378
SanDisk Ultra II	$63	$90	$173	$330
Transcend SSD370	$58	$90	$176	$359

The downside of being the slowest modern drive we have tested is that OCZ can only compete in price, but unfortunately the pricing isn't aggressive enough for the Trion to be competitive at all. Currently the ARC 100 is even cheaper than the Trion 100, which just doesn't make any sense because even OCZ is positioning the ARC 100 higher on paper and it's undoubtedly a better drive all way around. There is usually some level of decline from the initial street price soon after the launch, but in all honesty OCZ needs to cut the prices by 15-20% for the Trion to have a place on the market. OCZ told us that it will be running promotions with discounts, but it remains to be seen if those can bring the prices down to a level where performance and price meet. It has the lowest performance, thus it needs to be priced lower than the competition to provide the value to the user because right now I wouldn't have to think twice about buying the BX100 or 850 EVO over the Trion 100.

All in all, getting TLC done right is far from an easy job as we have witnessed with here. I believe that with the first big wave of TLC SSDs coming this year we are going to see sub-par performance compared to MLC drives. Fundamentally I have no problem with that because even a "slow" modern TLC SSD is a significant upgrade from a hard drive, but it is time for the manufacturers to realize that the price should reflect performance. It's just silly to take up to 50% hit in performance and only offer a few dollar savings because any educated buyer will gladly pay the extra few dollars for a substantially better drive. Once other NAND vendors start to ship 3D NAND in volume next year, we will likely see the majority of client SSDs move to TLC because as Samsung showed with the 850 EVO, 3D TLC NAND can enable planar MLC-like performance, but in the meantime it seems like MLC SSDs will still provide better overall value.