Overview

If AMD's HD 7970 debut Southern Islands card arrived in a fancy tux heralding a bunch of world firsts (first PCIe 3.0 card, first DirectX 11.1-compatible), this HD 7750 turns up to little fanfare in a Burton polo shirt and trainers.

The new Graphics Core Next's architecture has already been shown off by the HD 7970 and those 4.3 billion transistors pack quite a punch, as it turns out, trouncing the very best of last generations' GPUs by around 20-30% at mega-high res.

The HD 7970 is also excruciatingly pricey though. At £440 its staggering performance and overclocking capability are out of reach to most gamers.

The HD 7750 should arrive hitting the right side of £80, making it an altogether friendlier proposal, and these new-gen AMD cards boast some excellent power efficiency by shutting off all but one core when your system enters power save mode.

But what's this HD 7750 missing out on to hit that price point? Does it still make high-res screens sing?

Benchmarks

We put the HD 7750 through its paces in the super-taxing Heaven 2.5 benchmark at a gigantic 2560 x 1600 and stressed every nanometre of its circuitry in the most demanding DX11 games at 1920 x 1080.

The HD 7750 bested Nvidia's rival sub-£100 card, the GTX 550 Ti, but came a cropper against the HD 6770.

DirectX 11 tessellation performance: Heaven 2.5: Frames per second: Higher is better

AMD HD 7750: 9.4
AMD HD 6770: 8.0
EVGA GTX 550 Ti: 8.9

DirectX 11 gaming performance: Frames per second: Higher is better

DiRT 3

AMD HD 7750: 28.64
AMD HD 6770: 36.83
AMD HD 6850: 42.62

Shogun 2

AMD HD 7750: 20.33
AMD HD 6770: 25.85
AMD HD 6850: 32.0

Metro 2033

AMD HD 7750: 11.67
AMD HD 6770: 10.33
AMD HD 6850: 15.67

Just Cause 2

AMD HD 7750: 26.13
AMD HD 6770: 26.19
AMD HD 6850: 34.14

Verdict

Our opinions on this freshman 'Cape Verde' chip are dependent on UK pricing. If the $95 US price converts somewhere near the actual exchange rate without too much mark-up, it'll be available at £80 or less, and if that's the case you'll be getting some great performance returns for your outlay if you skipped a generation or two in your graphics card upgrade schedule.

If that UK price falls closer to the £100 mark though, it'll be a misfire, rather than one of those classic AMD bargains you wait for with each new gen's advent. The HD 7750 is quicker than its big Nvidia rival, the GTX 550 Ti, and its predecessor, the HD 5770 - but not the HD 6770.

General performance is limited primarily by a slender 128-bit frame buffer, however the die-shrink down from 45nm to 28nm and increase in transistor count that comes with it gives this Southern Islands card a definite edge in tessellation-heavy tasks. It's apparent in its strong Heaven 2.5 score, but in non-synthetic benchmarks the performance gain from the new architecture isn't as noticeable.

Overclocking the HD 7750's a mixed bag, too. On the one hand, it handles big core and memory clock increases smoothly and without crashes - we had ours cranked up to 900 MHz on the core clock from the 800 MHz stock setting without any glitching or hangs.

The downside though, is that we didn't really achieve a whole lot of performance increase by doing so – it only reported a 0.1 FPS increase when we ran the Heaven 2.5 benchmark. That, along with general performance, is likely to change as the 7700 series' drivers mature though.

We liked

We'll get a better idea of the HD 7750's true worth when firm UK prices arrive, AMD's drivers optimise and third parties get their overclocking mitts on it, but at present all evidence points to a strong budget card that'll appeal to anyone playing at 1080p or lower, and keeps an eye on power efficiency.

We disliked

We're hoping there's more potential under the bonnet for overclocking performance than we were able to extract with the early drivers, since AMD's flagship 7-series card the HD 7970 overclocked so damn well.

If it doesn't work out that way, this card still offers enough stock performance over its peers to make it worth a look.

Overview

AMD showed its hand first in this year's GPU arms race with Nvidia... by turning it into last year's arms race.

While Nvidia has kept shtum about its upcoming new 'Kepler' architecture and looks to do so until Spring, AMD stole the march and released the first of its new 7-series cards, the AMD HD 7970, a few days before Christmas 2011.

That sure was odd timing, but it taught us a lot about AMD's new Southern Islands architecture, specifically the 'Tahiti' chip. It's fully PCIe 3.0-supported, uses a 28nm manufacturing process to pack more transistors onto a PCB than ever before, and apart from offering very quick DX11 game frame rates, it's a highly energy efficient beast.

When your system drifts off into standby, the Tahiti card switches itself all but off too, minimising power draw.

The HD 7970's whopping £440 price made all those neat features all but irrelevant to the gaming masses though, so we're putting our hopes on this HD 7770 to deliver the best bits of the new AMD architecture for a more palatable price.

The 'Cape Verde' chip that this HD 7770 is built around makes full use of the new Graphics Core Next architecture like its big bro, with ZeroCore power efficiency in tow and a solid 1,000 MHz core clock. If it can offer a slice of the HD 7970's performance for this price, it's on to a winner.

Architecture, technology and specifications

The Graphics Core Next architecture is a bold move from AMD in that it represents a move away from the VLIW instruction used in 6-series cards towards a GPU-processing-friendly SIMD vector processor.

The previous processor type was great for graphics processing, but not suited to general purpose GPU computing - AMD left that side of things up to Nvidia and its CUDA cores.

Graphics Core Next is a u-turn on that philosophy though. GCN allows up to 16 data elements to be processed in a single clock cycle.

Grouping data before it runs through the vector processor is really efficient when dealing with general processing tasks - but the bad news for games is that you won't notice that difference in Battlefield 3 - the strengths of this architecture are wider-reaching than that, even as far as the professional market.

GCN also understands advanced languages like C++, meaning that in the long run, it'll be easier for developers to make use of the 7-series cards for complex programs.

The performance improvement from this architecture comes from passing data through a ton of compute units, which all work on the same operation until it's completed, and the resulting compute performance of this HD 7770 card is impressive at 1.28 TFLOPS.

It's built with ten compute units rather than the HD 7970's sixteen, but that's still enough to demonstrate a marked performance increase on last generation's equivalent model. And with AMD and Nvidia now adopting similar stances in their design, it's becoming an increasingly straight battle between the two - no hiding behind the blurred lines of CUDA cores and stream processors.

So the number of compute units and the simplified SIMD instructions they perform give AMD's 7-series cards the brains, but the clock speed is still the brawn of the operation. And at a world-first 1,000 MHz, it's fair to say the HD 7770 has brawn in check.

Benchmarks

We tested the HD 7770 with the most demanding DX11 around at 1080p, and with the Heaven 2.5 benchmark at 2560 x 1600 to really test its limits. It fared reasonably, demonstrating why it's more expensive than the HD 7750 and keeping in the same ball park as the pricier but older HD 6850.

DirectX 11 tessellation performance: Frames per second: Higher is betterHeaven 2.5

AMD HD 7770: 11.9
AMD HD 6850: 12.5
EVGA GTX 550 Ti: 8.9

DirectX 11 gaming performance: Frames per second: Higher is better

DiRT 3

AMD HD 7770: 41.73
AMD HD 7750: 28.64
AMD HD 6850: 42.62

Shogun 2

AMD HD 7770: 30.77
AMD HD 7750: 20.33
AMD HD 6850: 31.86

Metro 2033

AMD HD 7770: 13.00
AMD HD 7750: 11.67
AMD HD 6850: 15.67

Just Cause 2

AMD HD 7770: 31.18
AMD HD 7750: 26.13
AMD HD 6850: 34.14

Verdict

Many of the HD 7770's new features won't be noticeable right away. The underlying architecture is a big step forward for AMD that programmers and developers will find attractive, but for gamers looking to get playable frame rates out of DX11 games at 1080p its benefits aren't as obvious.

We're expecting the HD 7770 to enter at around the £100 point, which means it's going up against Nvidia's 550 Ti. While we didn't see it outperform the Nvidia card by the 100% AMD implied, it does hold a clear performance advantage, in addition to the subtler features under its bonnet.

Our sticking point though is that it was outclassed in every benchmark we ran by the previous generations' darling; the HD 6850. With just £20 difference between the two cards, we'd have liked to have seen a closer battle between the two.

It's worth mentioning that AMD's drivers tend to get better results from any given card after a few months of refinement, so further down the line we might see more parity in the benchmark figures between the two cards.

As with the HD 7750, the HD 7770 didn't impress in its overclocking performance the way AMD's high-end HD 7970 did. That £440 card maxed out AMD's overclocking software settings before crashing; the same can't be said here.

We managed to add on another 50 MHz to the core and memory clocks which yielded a modest increase in our benchmark tests, but it wasn't a reliable runner with those settings - glitches and crashes kicked in after a few minutes. Hopefully we'll see the full extent of the HD 7770's potential when the third party manufacturers release their beefed-up versions.

The 7-series might not all deliver staggering performance and overclocking, and that's to be expected given their wildly different pricings, but there is one feature that consistently impresses across the range - ZeroCore.

This is power-efficiency taken to a happy extreme – when your system enters a long-idle state, the HD 7700 along with all 7-series cards completely powers down the fan, 3D engine, compute units, shaders – virtually the whole card. The only activity going on comes from a small bus control block, which simply lets your computer know that the card still exists and no-one stole the GPU while it's been asleep.

We found it to be fast-acting, and dramatically reduced power draw when our system went idle.

We liked

This HD 7770 brings AMD's new architectural features to the budget audience with reasonable success.

Its GCN design and ZeroCore power efficiency make it a compelling argument to choose new over old, but it doesn't quite blow the best of the 6-series cards out of the water in plain old gaming performance.

We disliked

The HD 6850 can't do the 7-series cards' tricks, but it can render Metro 2033 frames quicker, and that makes the HD 7770 a tricky one to recommend.

As a stock card, it doesn't offer anything outstanding, but aftermarket companies might yet yield some impressive performances out of it.

Overview

The 120GB could be the SSD sweet-spot, but can Intel's SSD 520 Series 120GB drive hit that head-on?

Intel has done the obvious thing and stuck a SandForce controller in its desktop SSDs.

If the larger members of the new 520 Series, like the Intel SSD 520 240GB, are a little punitive on pricing, what about the Intel SSD 520 Series 120GB, could it offer the best compromise between price, performance and capacity?

One thing it certainly shares with the larger 520 Series solid state drive is Intel's SSD firmware development and validation regime.

It will take time for the true long-term performance of this latest Intel SSD family to shake out, but Intel has a well-earned reputation in this area.

The only slight snag is that smaller drives make for fewer memory chips and in turn fewer memory channels and a little less performance.

The optimal drive for performance in the range is the Intel SSD 520 Series 240GB model. But that's a much more expensive drive.

Instead this 120GB drive must do battle with the likes of the Corsair Force GT 120GB and OCZ's Vertex 3 120GB.

Benchmarks

Long term performance is the big worry with SSDs.

Early drives delivered scorching performance out of the box, but quickly went down the pan with intensive usage. To simulate a used drive, we install Windows 7 and then stuff each SSD full of data.

This used to be enough to make SSDs suffer slow downs. More recent SSDs are much more resilient and Intel's SSDs in particular have a strong reputation for excellent longevity.

As our benchmarks show, however, it's not hugely remarkable in terms of raw performance

Synthetic drive performance, compressible data

Synthetic drive performance, incompressible data

Synthetic drive performance, 4K Random

Application performance

Verdict

Check this out. Somewhere out there on the internet lives a species of PC hardware enthusiast that gets a kick out of endurance testing SSDs.

Among this strange breed, Intel SSDs have a very high reputation.

In fact, there have been reports of a 40GB Intel SSD soaking up 700TB of writes before finally losing the will to store data. When you consider that smaller drives fail faster in such conditions, well, it's pretty impressive.

So Intel's reputation when it comes to developing SSD firmwares that deliver excellent longevity is well earned.

Intel stresses that the same applies to the new 520 Series and its widely used, and strong-performing, SandForce SF-2281 controller. The firmware has been given the full Intel treatment.

That bodes very well for the long term.

What Intel hasn't done, however, is deliver a drive that immediately takes down the opposition in terms of raw performance. Most of our benchmark results are pretty much exactly what you would expect from a 120GB SandForce drive.

That said, there are a couple of areas where Intel's work might just be visible.

The Intel SSD 520 120GB is noticeably, if not dramatically, quicker than the Corsair Force GT 120GB in our random read and file decompression tests.

Unfortunately, sequential incompressible data write performance is no better than that same Corsair drive. Which means it's slower than both larger SandForce-based drives and the competition packing Marvell and Indilinx controllers.

We liked

The combination of Intel's attention to detail when it comes to SSD firware development and the sheer speed of the latest SandForce SF-2281 controller chip is pretty compelling. We also think 120GB is probably the current sweet spot in terms of balancing price with performance and capacity.

We disliked

Much cheaper than the 240GB model this 120GB drive may be, but it's still a significant investment to make. So it's a little disappointing that you not only have to give up half the capacity but also see write performance drop off, especially when shunting incompressible data around.

Verdict

SandForce performance plus Intel quality. Performance though is a little down on the 240GB version.

Overview

Intel is looking to SandForce to provide it with a performance-class SSD in the Intel SSD 520 Series 240GB.

It changed tack about a year ago and started using third-party controller chips in its consumer class solid state drives (SSDs). From that moment on, it was almost inevitable that a SandForce-powered Intel drive would eventually appear.

With the launch of the new Intel SSD 520 Series 240GB, and the other members of the 520 family, that day has finally come.

Of course, SandForce SSD controllers aren't perfect.

Thanks to the use of aggressive compression technology, impressive headline data transfer rates can sometimes translate into slightly pedestrian real-world performance. What's more, SandForce's second generation of controller chips suffered from a few stability glitches early on.

Still, there's no doubting the popularity, success and strong, all round performance of SSDs based on the latest SandForce controller. Add in Intel's reputation for going the extra mile with SSD firmware quality control and validation and you have a very promising mix.

The task for the Intel SSD 520 Series 240GB is clear enough. Not only must it rise above the hordes of SandForce based drives, it must also take on the new pretender in the form of the OCZ Octane and its refreshed Indilinx controller.

Game on.

Benchmarks

SSD performance is more multi-discipline decathlon than single-distance time trial.

A drive that delivers impressive peak performance doesn't always deliver the goods in random access workloads, for instance. Intel says it's put a lot of work into tuning the SandForce SF-2281 controller to improve performance and reliability.

But the results aren't always obvious in our benchmark tests.

Synthetic drive performance, compressible data

Synthetic drive performance, incompressible data

Synthetic drive performance, 4K Random

Application performance

Verdict

For the most part, Intel's new desktop SSD looks like any other drive based on the SandForce SF2281 controller chipset.

Not that we're suggesting that's a bad thing.

Sequential read and write speeds of 550MB/s and 520MB/s courtesy of a SATA 6Gbps interface is about as good as it gets for a single desktop SSD.

A peak IOPS rating of 80,000 for writes is pretty much par for the second-gen SandForce course, too.

So Intel isn't making any showbiz claims for basic performance.

In fact, if anything it's more up front about the limitations of the drive when it comes to shunting incompressible data around. In practice that includes most really big files, such as video, music and images.

Intel is also happy to 'fess up to the fact that the peak IOPS speed relates to a completely box-fresh drive. It only rates the Intel SSD 520 Series 240GB at a maximum of 60,000 IOPS in normal use.

However, what Intel does have is an enviable reputation for exhaustive validation and quality control when it comes to SSDs.

In fact, it's rumoured the 520 Series has arrived later than expected due to that very validation process. The upshot of all this, however, is not a drive than immediately blows away the competition for raw performance.

All our synthetic tests reveal an SSD that's very similar in performance to other SandForce-based drives. So that means a generally pretty fantastic performance with the exception of slightly ordinary incompressible data write performance.

We liked

It's not perfect, but SandForce's SF-2281 SSD controller is probably our current favourite.

It routinely wins a lot of benchmarks and is never too far behind even when it isn't at its best.

Add in Intel's hardcore validation and you have the promise of speed combined with longevity.

We disliked

If you were expecting something special in terms of raw performance, you'll be disappointed.

The 520 Series is very similar to other SandForce powered SSDs. And that means only ordinary performance in incompressible writes.

Like any large SSD, it's pricey, too.

Verdict

Probably the pick of the currently available 250GB-ish SSDs. Just a shame it's not a bit cheaper.

We looked at one of Plextor's first 6Gb/s rated drives - the PX-128M2S - last year. And barely nine months later, the next generation is upon us, in the shape of the M3 series. According to Plextor's own blurb, these are its most advanced drives yet.

The Plextor M3S SSD comes in 128GB (510MB/s read, 210MB/s write) and 256GB (510MB/s read, 360MB/s write) flavours, both of which are available now. The flagship M3S 512GB model (525MB/s read, 445MB/s write) will be appearing soon.

The new drives come with an up-rated Marvell controller, four times as much cache as the previous generation (in the 256GB and 512GB models), and faster NAND to complete the upgrade. So what do they look like when you throw some data down their tubes?

Marvell-ous

The new M3 series uses the same Marvell 88SS9174-BKK2 controller as the M2S series – and a load of other drives come to that – but with a twist: Plextor has made a raft of upgrades to the firmware. This, combined with 24nm Toshiba Toggle MLC NAND and a large 512MB cache (256MB in the 128GB model), gives the Plextor M3S 256GB pretty impressive performance.

Eight 32GB 24nm Toshiba Toggle NAND chips give the drive its capacity. These sit on the topside of the PCB, along with the two 256MB Nanya DDR3 1333 chips that make up the cache. The controller is positioned all on its lonesome on the underside of the board.

TechRadar Labs

Sequential read/write (incompressible)
AS SSD: MB/S: Bigger is better

Plextor PX-256M3S
Read: 468
Write: 346
Vertex 3 MAX IOPS
Read: 506
Write: 234
Samsung SSD 830
Read: 499
Write: 391

4K random read/write performance
AS SSD: MB/S: Bigger is better

Plextor PX-256M3S
Read: 24
Write: 46
Vertex 3 MAX IOPS
Read: 18
Write: 42
Samsung SSD 830
Read: 20
Write: 53

Under test conditions using the ATTO benchmark, the drive came pretty close to Plextor's own figures for sequential reads at 507MB/s, and did marginally better than the quoted figures for sequential writes at 365MB/s.

The more demanding AS SSD incompressible data benchmark saw it fall behind both the Vertex 3 Max IOPS and Samsung SSD 830 in the sequential read/write test, but it did much better in the 4K random read/write tests.

A bit of real world testing saw the drive take just four and half minutes to copy a 50GB folder of mixed file sizes and types, while loading Office 10 Pro took a smidgen under five minutes from key code entry to being ready for use.

Plextor is also shipping the M3S with its proprietary True Speed technology. This supposedly provides better sustained drive performance over time and, together with advanced wear levelling and bad block management, keeps the drive running closer to fresh-out-the-box than many of its competitors'.

The M3 series carries on from the previous generation, offering strong performance compared to the Samsung SSD 830 and OCZ's Vertex 3 Max IOPS. It's competitively priced as well, and at around £300 it may well ruffle a few feathers.

Plextor backed its previous range of SSDs with the usual three-year warranty, but the M3S series is backed for an unusually generous five years.

Verdict

Plextor's new drive is a rather impressive offering. It may not be the fastest kid on the block, but it's powerful enough and that price tag makes it a pretty attractive choice too.

Overview

Sapphire has waded in first with its take on AMD's latest HD 7000 series card, the Sapphire Radeon HD 7950 OverClock edition.

Unlike the AMD Radeon HD 7970 there are unlikely to be very many strictly reference models of the Radeon HD 7950. Instead we're more likely to see card manufacturers taking the opportunity to put their own spin on the PCB, clocks and cooling.

The HD 7970 was a reference model all the way down the line, with overclocked versions only coming much further down the release schedule. The Radeon HD 7950 though is being freed by AMD to allow manufacturers to do whatever they like right from day one.

With the impressive performance of the reference AMD Radeon HD 7950, which we've already seen, any improvement on that is going to be very welcome indeed.

And anything that offers even more overclocking potential with this already blazing fast GPU is just dandy in our books.

AMD's biggest card manufacturer, Sapphire, has opted to re-design the PCB and has created a new twin-fan cooling solution to sit atop the brand new slice of graphical silicon.

Sapphire has also opted to clock its Radeon HD 7950 OverClock edition 100MHz faster than the reference model we've already reviewed.

That's not quite as fast as the AMD Radeon HD 7970's 925MHz, but it's certainly not far short.

In fact, in certain areas, it's actually better.

Benchmarks

The benchmarks don't lie and show the Sapphire Radeon HD 7950 OverClock edition is only a fraction behind the vanilla Radeon HD 7970.

Thanks to the impressive overclocking potential within the Sapphire card, and the AMD Tahiti core in general, you can easily push the OverClock edition to the same performance levels as the HD 7970.

All of these benchmarks were run at 2560x1600, with the overclocking results taken at the relevant card's maximum achieved overclock.

There's little between them at the very top end of the OC tree.

DirectX 11 tessellation performance

DirectX 11 gaming performance

Overclocking performance

Verdict

As good as the reference version of the AMD Radeon HD 7950 is, the Sapphire HD 7950 OverClock edition is just better.

Coupled with the awesome overclocking potential of the Tahiti core is an excellent cooling array with whisper quiet dual fans.

Being able to keep the temperature well below that of the reference card gives us far more faith in the longevity of the chip given a serious overclock.

That's a good job considering the excellent Trixx overclocking utility Sapphire bundles with the card.

That software is intrinsic to the extra performance we were able to wring out of the HD 7950 OverClock edition.

The AMD overdrive software runs out of track at 1,100MHz, while Trixx will allow us to push the card up to 1,300MHz and play around with the voltages to help us get close to the limits.

We quickly managed to hit a 1,170MHz core clock speed with the Sapphire sample we've been playing with.

And that's faster than we were able to push the Radeon HD 7970.

So the performance is excellent, on par with the top-end AMD HD 7970, and the cooling array makes for near silent graphical power play.

There's got to be a catch, right?

Well, we're still looking, but we'll be damned if we can find one…

At £360 it's only a tenner more expensive than AMD is recommending the reference design be sold for, and around the same amount cheaper than the rivalling Nvidia top-end card.

We say rivalling, but really there is no contest. The Sapphire HD 7950 is cheaper and knocks the Nvidia GeForce GTX 580 about like a tiger playing with a toddler.

Nvidia then has some work to do to rival this excellent card.

The only problem we can see has no real bearing on the end user at all and is only really an issue for AMD.

Nobody is going to want the HD 7970 now there's a card which is practically as good for over £100 less.

We liked

The raw performance of the HD 7950 is a known quantity now, as is its overclocking potential. With the Sapphire backing though it's an even better GPU.

There's also the power-saving goodness of the AMD ZeroCore Power technology which turns off most of the GPU when it's not needed.

We disliked

We're struggling to think of something negative to say here. Aside from the fact you'll only get the true value of this card with a big-screen, 2560x1440 or 2560x1600 monitor, there's nothing to say.

Verdict

Quite simply this is the only card we'd consider spending cash on if we were looking for a serious GPU upgrade.

Overview

AMD is really putting the pressure on Nvidia now with its second release of the new AMD HD 7000 graphics card generation, the AMD Radeon HD 7950.

Nvidia is still sitting back waiting for the right moment to strike back, but can it recover from these two quick blows?

Well, we say quick - it's been well over a month since AMD launched its first card of this generation, the AMD Radeon HD 7970.

That was a surprise given that, pre-Christmas, we were expecting both cards to hit the streets at the same time in the first week of January with a possible dual-GPU iteration coming around now.

AMD though decided to give its top-of-the-line, £500 AMD HD 7000 card a bit of breathing space at the start of its life, and now that the AMD Radeon HD 7950 is sat here in our labs it's easy to see why its release was delayed.

Essentially it's almost as good a card for over £100 less.

So AMD's claims to be delaying so it could wait for AMD Radeon HD 7950 units to be in the market (despite launching its big brother, the AMD Radeon HD 7970 well before you could even lay eyes on one) seem to be rather thin.

We think it's more likely AMD realised even fewer people would pick up a £500 graphics card when there was one for £350 that could do the same job practically as well.

To be fair though this isn't the first time this has happened; the previous generation had exactly the same problem in the two top-end Cayman cards – the Radeon HD 6950 and Radeon HD 6970.

There was precious little difference between the two cards in benchmarking terms, and with some judicious use of BIOS tweakery and ROM flashing fun, there ended up being precious little difference between them architecturally too.

But there must be some differentiating factors, some reason for the £150 price difference.

So, what has AMD chopped out of the AMD Radeon HD 7950 Graphics Core Next GPU to make the grade?

Architecture

The AMD Radeon HD 7950 is still based on the exact same Graphics Core Next/Southern Islands architecture as its only other AMD HD 7000 sibling, the HD 7970.

That means it's a graphics card still nailing the very latest of technologies in its rather sizeable package.

It's a fully fledged DirectX 11.1 card (though that's not actually going to be around soon, or even that big of a deal), more of interest though is the production process, shrinking down from the 40nm of the Cayman GPU to 28nm.

That's smaller than the current crop of CPUs.

That also means it can pack in a hell of a lot of those teeny transistors into the GPU, and AMD hasn't stinted. It's thrown 4.3 billion of them into this Tahiti core.

But it is a slightly chopped version of the Tahiti XT powering the AMD Radeon HD 7970.

This Tahiti Pro comes with only 28 of the Compute Units that make up the new vector processor AMD is now using for its graphics cards.

That means it's only rocking 1,792 shaders/Radeon Cores/streaming processors, or whatever you want to call them. Compared to the 2,048 of the Tahiti XT core that's a bit of a drop.

It also loses out on some 16 texture units, but thankfully it's still got the same full complement of ROPs at 32.

Outside of the GPU itself, the card as a whole comes with the same huge 3GB of GDDR5 frame buffer, on the same 384-bit interface.

And all still on the burgeoning PCIe 3.0 technology.

The clocks have also, inevitably, been cut back too.

The stock HD 7950 comes out of the box at 800MHz, a very conservative setting compared to the 925MHz of the HD 7970.

So, the AMD Radeon HD 7950 is not actually missing out on too much of AMD's new tech, indeed it's still hitting over 3TFLOPs of processing power to the HD 7970's 3.8TFLOPs.

For gamers though, forget FLOPs, it's polygon-pushing performance in games that we really want to know about. So how does that 20% drop in raw processing power translate into gaming benchmarks?

Benchmarks

At stock speeds the difference between the AMD Radeon HD 7950 and HD 7970 is actually quite pronounced, sometimes by as much as 25%.

That's a pretty reasonable performance difference between your top two GPUs, but it's not actually much to do with what's been chopped out of the GPU architecturally, more to do with the drop in clockspeed.

Once things are evened out in the core clock, through overclocking, there is suddenly almost nothing between the two AMD HD 7000 series cards.

The CrossFire performance is impressive for the AMD Radeon HD 7950 too, offering similar frame rates to the pricier cards, and the savings over the HD 7970 are multiplied in multi-GPU arrays.

DirectX 11 tessellation performance

DirectX 11 gaming performance

DirectX 10 gaming performance

Overclocking performance

CrossFire performance

Performance

As with any graphics card release it's all about performance and the AMD HD 7950 has got it in spades, and with a fair amount in reserve too.

At stock speeds the pace the AMD HD 7970 sets is tough for the HD 7950 to follow, leaving it at least 10% slower than its big brother. Sometimes that gap widens leaving the HD 7950 around 20% behind.

That's something that we would normally expect between the top two cards of a manufacturer's new generation.

But what of our assurances the AMD Radeon HD 7950 was almost as good a card?

Well as soon as you start waving around that familiar overclocking stick you can quickly see much of the difference between the two cards was taken up with the separation in core clockspeed.

Of course there's no guarantee that every Tahiti Pro GPU will be capable of these feats of overclocking, but the yields of these chips are only going to improve so I wouldn't be in the least surprised to see every Tahiti Pro capable of topping the 1GHz mark.

Like the HD 7970 before it we were able to drop into AMD's Overdrive software and push all the clock and memory sliders to their maximum settings without a problem.

In the case of the AMD Radeon HD 7950 that meant we hit 1,100MHz on the core and 1,575MHz for the memory.

That's fast.

And, as you can see from the benchmarks on the previous page, when the clockspeeds are pushed to the same limits there is almost nothing between the two top AMD HD 7000 series cards.

Considering the HD 7950 is over £100 cheaper than the HD 7970, that's quite impressive.

Interestingly though you don't even need to overclock the cards to make that difference disappear in CrossFire.

When the HD 7000 series cards are paired up there is again very little to separate them in terms of performance, and nothing that couldn't be closed if you just pushed the HD 7950 up to 925MHz.

Verdict

It's almost a shame AMD released the HD 7970 first when it could have had an instant hit, and a lot of good feeling over a month ago with the AMD Radeon HD 7950.

Though obviously people wouldn't have been quite so inclined to spend out for a HD 7970 when the cheaper card was just as capable a pixel-pusher.

But still, none of that can take away from the fact that the AMD Radeon HD 7950 is an excellent graphics card.

It's not just the AMD in-fighting that places this card at the top of the current crop of graphics cards.

We've been pushing the Nvidia GeForce GTX 580 as the go-to gamer's card since it was released, but the HD 7950 has that beat and for a good chunk of cash less than the Nvidia card.

And that's just at stock speeds. When you start overclocking this card the difference in performance increases hugely.

Even if you've never overclocked a graphics card in your life you owe it to all the engineers who worked on the Tahiti GPU to push it north of the 1GHz mark.

The reference cooler design, with its vapour chamber technology, is easily capable of absorbing the extra heat so the chance of causing any damage in doing so is negligible.

And all you need to do is bring up the AMD driver panel and push a couple of sliders.

Job done, instant, awesome frame rates.

Inevitably there are a host of factory overclocked Radeon HD 7950s on their way, like the Sapphire Radeon HD 7950 OverClock Edition, which will allow you to push the GPU even further.

The AMD HD 7950 could also be a massive hit for the CrossFire crew too, as for £700 you'll find yourself with an insanely quick graphics setup.

And for £300 less than an equivalent HD 7970 array.

Speaking of CrossFire too, it's worth mentioning that, like its big brother, the HD 7950 comes with the impressive AMD ZeroCore power tech.

That means if you've got a pair of GPUs running in your machine they'll only both be drawing power when you run a game. In normal desktop mode the second GPU switches off entirely, reducing power draw and unnecessary fan noise.

The primary GPU will also shut itself off almost entirely when the rig's screen goes into standby too.

For such power hungry gaming rigs these power saving functions are vital, and very, very welcome.

So AMD has got itself a real winner here with the HD 7950, even if it will inevitably cannibalise the sales of the pricier HD 7970.

We liked

The stock performance of the HD 7950 is impressive, but it's the amount of head-room for overclocking that Tahiti Pro core represents which really makes this card.

The ZeroCore Power tech is another of our favourite things about the HD 7950. Realistically you'll never notice it in action, but it will be there, saving you money unobtrusively in the background.

We disliked

The only thing to dislike about the card is the fact the clockspeed has been set so artificially low. Thankfully though AMD hasn't locked the overclocking possibilities down as it did with the HD 6950.

We did try flashing the BIOS of our reference HD 7950 to see if we could re-enact the fun we had unlocking the dormant cores in the Cayman GPU with the HD 6950.

Sadly while we were able to boot with the HD 7970 BIOS on the HD 7950 it didn't unlock the extra 256 Radeon Cores. And it didn't help with stability either…

Verdict

The AMD Radeon HD 7950 is one hell of an impressive pixel-pusher, and Nvidia is going to have to work incredibly hard with its Kepler cards to best this excellent card.

The fast transfer speeds and high capacities of SDHC cards has made them ideal for DSLR camera stills and videos, and their price continues to fall. At around £26, the Samsung 16GB SDHC Plus Class 10 card may not be the cheapest currently available, but it justifies its premium by boasting a number of protective qualities in addition to high read and write speeds.

Samsung claims that the card is waterproof, even against a 24-hour stint in sea water, and that it can withstand both the pressure of a 1.6 tonne vehicle and strong magnetic forces. Even if the card is beyond repair, claims Samsung, its data may be retrieved, and with the standard lock tab it's possible to prevent data from being recorded or deleted.

Adhering to the Class 10 SD card specification, the card offers maximum read and write speeds of 24MB/s and 21MB/s respectively, which is more than reasonable, although slightly behind the 30MB speeds offered by Sandisk's equivalent Class 10 card.

Its metallic finish is sleek, and when the card is bent it stays noticeably more rigid than the casings of standard plastic SDHC cards.

Tested in a Nikon D3100 body, the card was pitted against a rival manufacturer's Class 10 SDHC card. While the camera managed to maintain the same number of frames and speed with each, photos appeared to be written from the buffer to the Samsung 16GB SDHC Plus Class 10 card at a slightly quicker pace, since subsequent images could be taken sooner. Against the same card, images were also transferred to a computer around 20% quicker.

Verdict

If you don't need its protection, you'd be advised to shop around, since you may be able to find similarly-priced cards with slightly faster speeds. But its performance shows that it's nevertheless a competitive option. It also appears as the ideal partner for the growing range of waterproof compact cameras, should anything go wrong.

Overview

Not to be outdone by new graphics releases AMD is going back to the APU drawing board with the new AMD A8-3870K processor.

While its flagship FX processors are failing to shine it does seem a little on unfair on AMD that at the other end of the market it has a chip which really ought be cleaning up.

Its Llano Fusion APUs, which combine a multi-core CPU and a Radeon graphics part on one die, are actually rather good.

They may be based on an older processor architecture, but quad core examples like this A8-3870 hold their own against Intel's similarly priced dual core Core i3s in CPU benchmarks.

When it comes to 3D tasks like gaming, the on board HD6550D is simply in a different league to Intel's laggardly HD Graphics 3000 cores.

How could they get better? We'd like a chunky price drop, but that's not going to happen.

Instead, AMD has introduced 'Black Editions' of the chips, recognisable by their 'K' suffix. Just like Intel chips with a K moniker, they come with an unlocked multiplier which makes them more suitable for overclocking.

Now, you might think that anyone who wants to overclock a chip will be looking at something a little more high-end than these very basic processors, but the entire raison d'etre for overclocking is to turn something cheap into an unexpected powerhouse.

An unlocked, gaming-capable hybrid chip for just over £100 sounds like the stuff hardware hackers' dreams are made of. But does it deliver on that promise?

Benchmarks

As expected, even with a 100MHz increase over the older AMD A8-3850, at stock speeds the A8-3870K doesn't distinguish itself fully against the Core i3.

Put the two chips head to head solus and it has the edge, but if you're using external graphics the Core i3 still pulls away in any game that's not massively multithreaded (Shogun 2, tellingly, is).

And what of the overclocking? It's actually a bit disappointing really. At the speeds we achieved using fairly basic equipment, the difference is often not enough to be worth even trying.

DirectX 11 gaming performance

DirectX 11 tessellation performance

CPU rendering performance

CPU video encode performance

Verdict

A stock A8-3870K ships with its CPU set at 3GHz and the graphics core clocked at 800MHz. Our attempts to increase these weren't exactly stellar - it ramped straight up to 3.4GHz/800MHz, but any attempts to go further were unstable enough to be frustrating.

With more fine-tuning and a really good cooler, we've seen reports of much higher clockspeeds, but we're not overly sure it's ultimately worth it for the performance returns.

Plus, it's unlikely that motherboard manufacturers are going to get behind this platform with high end boards full of good, stable overclocking features – because potential purchasers just aren't going to spend the extra they'll cost to make.

Still, since the A8-3870K is already available for less than its non-tweakable predecessor, the A8-3850, the unlocked multiplier is more of a pleasant extra to have.

So even if our dream of using one to outperform a Core i7 isn't going to happen, the A8-3870K is still a great little chip.

The CPU part of Llano was never in much doubt. Based on the same architecture as Athlon II and Phenom II with a die shrink, it's old but still capable.

Because it has four native cores it outperforms Core i3 in tasks like media rendering and so on, although Intel's computational engine is still superior in games – even if its graphics lags far behind.

For a workplace PC, then, the A8-3870K is an exceptional choice. It'll beat a similarly priced Core i3 system at everything.

It gets a bit more complicated if you're after a budget system that can game at 1080p – to use as a media centre, for example.

Even overclocked, the A8-3870K alone isn't an alternative to discrete graphics – although it does come very close if you're willing to sacrifice graphics quality. Battlefield 3, for example, is this far off being playable at 1920x1080 with low image quality settings.

Which are still bloody good looking.

Skyrim at similar settings runs like a dream.

If you're thinking of adding in a mid-range card like the HD 6870 or even something as powerful as an Nvidia GeForce GTX 560Ti, though, Intel just wins through – although it's a draw in CPU limited games like Shogun 2 and Skyrim.

AMD's architecture does have the curious disadvantage of performing slower in CPU benchmarks when discrete graphics are attached too.

But that leaves one feature that we haven't discussed yet which bring us down in AMD's favour.

That feature is Asymetrical Crossfire – the ability to use both the A8-3870K's graphics and a low power GPU, such as an AMD HD 6670, for a gaming experience that's capable of playing most games at 1080p with medium settings.

That's a hell of a thing in AMD's favour.

Enough that while it's not going to be our chip of choice for a workstation or enthusiast games rig, if you want a small, low cost PC which is capable of occasional games at console quality, it's a steal.

We liked

Where it comes into its own is as a flexible platform for a quad core office machine or a flexible and capable gamer with Asymmetric Crossfire.

We disliked

The potential to overclock is a bonus, but really not enough by itself to make us favour this chip over any other quad core Llano.

Final word

Not the budget buster we hoped for, but further's our faith in AMD to make Fusion the low cost platform of choice.

Overview

No you haven't woken up in a strange parallel universe where everybody's making everybody else's products. And yes, the Zalman SSD-F1 Series 240GB is made by the same Zalman that makes all those fancy-looking CPU coolers, among other things.

In fact this actually isn't Zalman's first go at a solid state drive (SSD) range - it had two drive ranges out previously, the Zalman N series using a first generation SandForce controller and the S series using a JMicron chip.

Now it has a new family of 2.5-inch SSDs - the F1 Series - to enable it to play with the big boys without sand being kicked in its face. This new drive uses the SandForce SF2281 controller and SATA 6Gb/s interface, and should make for some interesting comparisons with drives that have been around since the launch of the latest range of SandForce controllers.

This 240GB drive is the current flagship of the range, quoting IOPS (4K random) read and write speeds of 45,000, sequential reads of up to 560MB/s and sequential writes of up to 530MB/s. There are also 60GB and 120GB capacities available, both with quoted IOPS (4K random) read/writes of 30,000.

Benchmarks

To test SSDs in as close to a real life scenario as possible, the drive was installed as a boot drive and a full operating system was installed and motherboard drivers loaded. We tested in this state, then filled it up with data, then deleted it, running through this cycle a few times before testing the drive again to see if there was much difference in the results after subjecting the drive to this usage cycle.

Sequential read/write performance (compressible data)

Sequential read/write performance (incompressible data)

4K random read/write performance

Verdict

With everyone and their dog seemingly using the second generation of SandForce controllers as a basis for an SSD or three, there is always a tendency to yawn a little and offer a terse "go on then, show us what you've got."

Well, in the case of the Zalman SSD-F1 Series 240GB, that happens to be quite a lot. The drive is blessed with stunning performance and is more than a match for some of its SandForce competition.

While distracting the outside world with pretty-looking and highly efficient coolers, some people deep inside Korea were keeping a watchful eye on what was happening with the SSD market and biding their time before launching this second generation drive.

It appears that Zalman has taken full advantage of this time by the way the drive can keep up with the OCZ Vertex 3 MAX IOPS in our benchmarks.

And it also shows it has access to the latest firmware tweaks that were applied to the MAX IOPS, despite the Zalman SSD-F1 Series 240GB drive using the Intel's 25nm MLC flash memory of the original Vertex 3, not the 32nm Toggle NAND that the MAX IOPS version uses.

Benchmark figures are all well and good, but what about in the real world, what does the F1 240GB drive bring to the table?

Well, installing Windows 7 from scratch took a mere 16 minutes from start to the password entry screen, while the operating system itself takes just 36 seconds from a cold boot to be ready for use. Loading a full copy of Windows Office 2010 Pro took a very rapid 4 minutes 41 seconds.

We liked

The Zalman SSD-F1 Series 240GB is a real shot from left field; no one really saw it coming, but performance-wise it really is hard to fault. It's not that badly priced either. Yes it's a bit more expensive than some of the competition but, perhaps more importantly, it's cheaper than the Vertex 3 MAX IOPS, which it practically matches for performance.

We disliked

Having to download the migration software that comes in the shape of Acronis True Image HD is a bit of a pain, and there's no real excuse for it not being part of the box bundle. But no doubt Zalman would say it's to keep the cost down. But hey, that's a teeny logistical issue, not an actual SSD problem.

Final verdict

From nowhere to be among the leaders of the pack of SSDs is an impressive achievement from Zalman, to say the least.