Titan Fenrir Evo Hands-on

Posted on 17th May 2010 at 11:40 by Richard Swinburne with 48 comments

Once upon a time, when PC games used to be 2-dimensional, every kind of graphics processing was done on the CPU. This was also true for early 3D games. Neither Wolfenstein 3D nor Doom with its numerous clones listed a graphics accelerator among their system requirements. Well, they couldn’t since there were no graphics accelerators at that time and game developers did not rely on them. Moreover, talking about gaming applications of their products, Intel and AMD focused on enhancing the capabilities of their CPUs in the way of MMX and 3DNow! instruction sets. Intel promoted MMX as a means to boost the quality, level of detail and speed of gaming graphics whereas the name of AMD’s technology speaks for itself.

This situation went on for quite a while. Even rather late projects by id Software and Epic Games such as Quake and Unreal used the CPU as the main tool for processing graphics, notwithstanding the significantly expanded game worlds. Things changed in 1996 when the obscure and young firm 3dfx Interactive unveiled the world’s first 3D graphics accelerator for the PC affordable for ordinary gamers. The product looks ridiculous by today’s standards as it could only map and filter textures, but the quality of the filtering was unprecedented at that time. None of then-existing CPUs, whatever multimedia instruction sets they supported, could deliver such performance even at a lower image quality. A game would look completely different running in the Glide mode as opposed to software mode. 

That was the first revolution in the world of gaming 3D graphics. CPUs were losing their ground year by year, their influence on performance constantly diminishing. There were other turning points, the next one being the Nvidia GeForce 256, the world’s first graphics processor with a TCL unit (Transformation, Clipping, Lighting) that could transform 3D coordinates into 2D ones, clip polygons and light the scene, offloading the CPU. As is often the case, the new product did not take off immediately. Nvidia’s opponents still relied on the growing computing capacities of CPUs for those tasks, yet hardware TCL had become widespread by the end of 2001 anyway. The same year there was a third revolution that expanded the capabilities of GPUs even more by making them programmable. The Nvidia NV20 (GeForce 3) came out as the first chip to support DirectX 8.0. And the last notable innovation occurred in 2002 when ATI Technologies announced the R300, the first GPU to support DirectX 9.0.

From that moment onwards, GPUs were developing in an evolutionary way. New versions of DirectX and OpenGL were being implemented. The computing part, originally divided into vertex and pixel processors, became unified. New types of shaders were supported and there were lots of other innovations. GPUs quickly surpassed ordinary CPUs in sheer computing power, giving birth to the idea to use them not only to process graphics but also to accelerate complex computations unrelated or loosely related to 3D applications. Both leading developers, AMD and Nvidia, are working actively in this direction, but that’s not the point of this review. Looking at the computing capabilities of today’s GPUs estimated at teraflops (more than the huge supercomputers of earlier times could offer!), one might find modern CPUs to have rather humble parameters.

This provokes a natural question if 3D games need powerful CPUs at all. The answer is not as simple as it seems. First, if some GPUs resources are allotted to compute the game AI or physical model, there are fewer resources left for graphics processing. And we know just too well that today’s games have very complex visuals that may take all the 1600 stream processors of an RV870 chip to be rendered at a decent frame rate. Second, it is not so easy to rewrite the game code to make maximum use of GPU resources. It looks like a number of computational tasks, including gaming ones, are still performed better on the CPU, therefore premium-class gaming computers from famous manufacturers like Alienware are equipped with extremely fast and expensive CPUs. Particularly, the quad-core Intel Core i7-975 Extreme Edition cost $999 when announced and the newest six-core Core i7-980X is going to cost that much today. That’s quite a lot, but a top-end graphics card is expensive as well. For example, a Radeon HD 5850 costs about $300 whereas topmost solutions that deliver maximum performance are as expensive as $600-800 (a dual-processor Radeon HD 5970) or even $1000 and more (a couple of Nvidia GeForce GTX 480 cards working in SLI mode).

Sapphire Radeon HD 5970 4GB Toxic Review

Manufacturer: Sapphire
UK Price (as reviewed): £900+ (inc. VAT) MSRP
US Price (as reviewed): $1099 (exc.Tax) MSRP

Pre-overclocked graphics cards have been around for years and, while not for everyone, can offer that extra sliver of performance if you’re willing to foot the bill. What we regularly ask though is why manufacturers hold back so much? If our own experiences, many graphics cards yield decent amounts of overclocking headroom, yet pre-overclocked cards often settle for tiny overclocks of 3-5 per cent. Even watercooled pre-overclocked cards rarely stray beyond overclocks of 15 per cent.

Part of the reason is the availability and quality of individual GPUs and their ability to handle higher clock speeds, but Sapphire has taken note, and decided to really push the boat out with its new Radeon HD 5970 Toxic. Not only do both GPUs ship with a massive 24 per cent overclock, but the card comes fitted with twice the GDDR5 memory of the stock model, and a massively upgraded cooler too. Make no mistake, this is a pre-overclocked partner card done to the limits of what’s possible, to the point that the production run will be extremely limited; Sapphire tell us that less than one thousand units will be produced.

Underneath the pomp, ceremony and huge cooler (more on that in a bit) though, this is still a Radeon HD 5970 card, already the most powerful graphics card on the planet even before Sapphire started tinkering. The Toxic is still 31cm (12.2in) long and is still built around two full-fat Cypress GPUs (codenamed Hemlock XT).

After Sapphire’s tweaking and the addition of the massive cooler, these GPUs run at 900MHz, up from the 725MHz of a stock-speed HD 5970. This theoretically makes the Toxic comfortably faster than two HD 5870s (which share the same GPU and are clocked at 850MHz) in CrossFire. To achieve these clock speeds, Sapphire has had to increase the GPU’s voltage, and this has necessitated the two 8-pin PCI-E power connectors used to fuel this beast. Sapphire recommends an 850W PSU for any PC that includes the Toxic.

The doubling of memory from 2GB on the stock HD 5970 to 4GB also requires extra power. This is doubly true, as Sapphire has chosen to use twice the number of DRAM chips, rather than use higher density 256MB modules, with 16 chips on either side of the PCB. This has required a slight PCB reshuffle, as well as extra power phases compared to the typical HD 5970, but the stock HD 5970’s PCB is roomy enough to allow for the extra memory without issue. The memory has also been overclocked, from 1GHz (4GHz effective) to HD 5870 speeds of 1.2GHz (4.8GHz effective).

With the extra power coursing through its circuitry, not to mention the noise that a stock HD 5970 makes under load, replacing ATI’s stock cooler was a mandatory requirement for the Toxic – this is a likely reason why we’ve not seen many attempts at a pre-overclocked HD 5970 yet. Sapphire’s cooler is a monster fit for this purpose though; it’s a modified Arctic Cooling Accelero Xtreme H5970 triple-slot construction of heatpipes, aluminium fins and three slim-line 92mm cooling fans. We’d recommend leaving another slots-worth of space beyond this to allow the fans to work efficiently.

Each GPU is fitted with a copper contact plate through which run four nickel-plated copper heatpipes. These two sets of bi-directional heatpipes dump their heat into the cooling fins of the cooler, with the cooling fans blowing directly down through the fins and onto the PCB. The card also uses a backplate to cool the 2GB of GDDR5 memory mounted there. Due to the upped power demands of the card, there’s also a separate VRM heatsink that sits above the card’s power delivery circuitry, and which is secured by both thermally conductive tape, and screw-through bolts.

At the rear of the card are the usual pair of DVI ports and the mini-DisplayPort output, and the card supports three-screen Eyefinity. The Toxic is certainly a very impressive card on paper – superior to two HD 5870, with the advantage of having the two GPUs closer and fed by a PCI-E bridge chip, plus the significantly upgraded cooling. We were understandably eager to see how this monster of a graphics card performed.

Specifications

• Graphics processor 2 x ATI Radeon HD 5970, 900MHz
• Pipeline 2 x 1,600 stream processors (900MHz), 2 x 32 ROPs
• Memory 2 x 2GB GDDR5, 4.8GHz effective
• Bandwidth 2 x 153.6GB/sec, 256-bit interface
• Compatibility DirectX 11, OpenGL 3.1
• Outputs/Inputs 2 x DVI, 1 x mini-DisplayPort, 1 x CrossFire
• Power Connections 2 x 8-pin, side-mounted
• Size 310mm long, triple-slot
• Warranty two-year

Do you know what is peculiar about EVGA mainboards based on Intel H5x series chipsets? There may be different answers but I personally find it strange that there is no H57-based product among them. There is a section called “Intel H57/H55 Series Family” at the EVGA website but it only includes two mainboards based on the H55 Express. That’s odd, isn’t it? In fact, EVGA announced three mainboards, one of which was based on H57, when Intel unveiled its new chipset series. There is still some information you can find about the product with the unassuming name of EVGA H57. It is a full-size mainboard with two PCI Express x16 slots and with a couple of extra controllers, one of which adds PATA and two SATA ports and another, two IEEE1394 (FireWire) ports. That mainboard is almost a copy of the product I am going to talk about in this review, the EVGA H55, except for the differences coming from the use of different chipsets. For example, EVGA H57 has two USB ports more and also has two PCI Express x1 slots.

So, this review is about the EVGA H55, but the company also offers another H55-based mainboard called EVGA H55V. It is a microATX product with a simpler layout. There are no additional onboard controllers; the components of its voltage regulator circuitry are not covered with any heatsinks. The ATX12V power connector is 4-pin and there are numerous electrolytic capacitors next to the solid-state ones. This mainboard has some advantages over the other two models, though. Particularly, there is an eSATA port at its back panel.

By the way, you can spot a place reserved for an onboard pin-connector for two more USB ports on the EVGA H55V. It means that EVGA may release a mainboard (called something like EVGA H57V) with the same PCB design and with the H57 chipset.

I don’t know exactly why EVGA has decided to give up H57-based products, particularly the EVGA H57. But let us take a look at the similarly designed EVGA H55 and check out its features. Perhaps we will find the answer.

How AMD Core Unlocking Works

We’ve spent the last couple of weeks travelling the length of Taipei, seeking answers from the three main motherboard companies: ASUS, Gigabyte and MSI, learning how exactly they’ve each achieved the core unlocking capacity for AMD AM3 CPUs, especially as it’s been officially removed by AMD from its latest 8-series chipsets.

We’ll start with a bit of background first. At the time of AM2+’s demise, AMD was losing ground (and money) hand over fist to Intel and needed any advantage it could grab. While AMD must have been concerned about the fact that core unlocking could erode the market for the company’s latest Phenom II X4 CPUs and pushed the ASP (average selling price) downwards, the vast majority of consumer CPUs it sells are sub-£100 models and core unlocking helped revitalise interest in these.

With its previous SB710/SB750 Southbridge, AMD introduced a new feature called Advanced Clock Calibration. This included an “EC Firmware” function in the BIOS that could be issued a set of codes (subsequently provided by AMD to partners) and this would enable the extra cores of compatible CPUs that had previously been disabled. The EC Firmware is specific to each CPU that has a hidden core(s), so there’s a list of codes need for all the CPU individually, rather than just one global setting.

In the latest SB850 southbridge AMD hid the EC Firmware function and declined to issue a new set of compatible codes, much to the dismay of motherboard manufacturers. They could either accept that it was game over for core unlocking, or call in the engineers and come up with their own ways to get it working again. They feared that without it – and facing a rejuvenated £100 Intel line-up with the excellent Core i3-530, interest in AM3 would decline.

Asus

First out the blocks with an 890 chipset board capable of CPU core unlocking was Asus’ 890GX based Asus M4A89GTD Pro/USB. Clearly it took the other motherboard companies by surprise as both MSI and Gigabyte were at least a month behind. It’s highly unusual for one company to have such a big, clear lead – chalk one up to Asus’ secrecy there. It wasn’t until recently that Asus would let us in on how it able to achieve this.

Asus engineers basically replaced the whole advanced clock calibration circuit (ACC) and then substituted in a hardware switch which issued its own EC Firmware codes for the CPUs instead. This costs a few dollars more, but Asus claims it not only gives more control to select individual cores (if there’s more than one available) for the maximum unlocking success rate, but given previous experience the engineers have developed a feature to test and check if the unlocked cores even function correctly and disable bad ones automatically.

On Most Asus boards this option is available with a BIOS switch, but with its recent [Crosshair IV Formula motherboard it now has a hardware button on the board to turn it on and off, although we’re unsure if it still requires the complete system reboot in the same way.

AMD used to make it very obvious that it had no interest in the market of miniature and low-power computers with limited performance, i.e. netbooks and nettops. Of course, this has not prevented some makers from offering such computers built with AMD components. For example, a compact desktop machine called Zino HD from Dell. However, this trend hasn’t yet become mainstream, and most nettops out there are still based on Intel and Nvidia hardware. Moreover, there are almost no AMD parts in the market that would allow any user to build a mini-ITX nettop on their own.

It isn’t AMD cannot produce CPUs with low heat dissipation that could be used as a basis for computers like that? On the contrary, AMD offers Socket AM3 processors for desktop PCs with a thermal design power of 45, 25 and even 20 watts. In our opinion, CPUs like that could become popular among people who would like to assemble a nettop with their own hands as they offer good performance, especially in comparison with Intel Atom, and are just as good as Intel’s modern LGA775 Celeron series in terms of heat dissipation. It looks like AMD’s power-efficient CPUs for desktop PCs are a good offer that fits perfectly into the free market segment between Intel Atom and Intel Celeron series.

It is a different story with the mainboards. AMD’s belief that their CPUs are no good for compact multimedia PCs has led to the fact that there are nearly no modern Socket AM3 mainboards in the mini-ITX form-factor in the today’s market. It is a shame considering that AMD’s new integrated chipsets such as the AMD 785G and 880G would be optimal for nettops as they feature a rather fast graphics core capable of hardware HD video acceleration, support all modern interfaces, and have modest heat dissipation. So, the only obstacle that a developer of compact Socket AM3 mainboards has to face is the rather large physical size of the CPU socket which should also have a rather large cooler retention mechanism around it. Besides, each chipset from AMD consists of two chips, making designing small PCBs even more complicated.

Fortunately, some developers do not find those difficulties insurmountable. Sapphire’s recent release of a mini-ITX mainboard for the new generation of energy-efficient Socket AM3 processors has become a good stimulus for writing this review. We are going to talk about energy-efficient AMD processors in terms of their suitability for compact and economical computers for home and office.

AMD Phenom II X6 1090T Black Edition

Manufacturer: AMD
UK Price (as reviewed): £240 (inc. VAT) (estimated)
US Price (as reviewed): TBC
Preferred Partner Price: £233.94 (inc VAT)

AMD may have already raced ahead of Intel when it comes to integrating more cores into its CPUs with the launch of its 12-core Magny-Cours workstation/servers CPUs, but it’s taken considerably longer to release a desktop processor with more than four cores.

However, while Intel’s first six-core desktop CPU, the Core i7-980X Extreme Edition is brutally fast, it’s also extortionately expensive, retailing at £840, more than three and a half times the price of the quad-core Core i7-930.

In contrast, AMD has taken a completely different approach with its first six-core desktop CPU, the Phenom II X6 1090T Black Edition – pricing it far more reasonably.

Thus, at £240 the Phenom II X6 1090T Black Edition is only £95 more expensive than the quad-core Phenom II X4 965 BE and only £10 more than the quad-core Core i7-930.

The X6 1090T BE is physically indistinguishable from the X4 965 BE as it’s also a Socket AM3 processor. This means it should work in any Socket AM3 or Socket AM2+ motherboard that can handle a 125W TDP CPU with a suitable BIOS update.

AMD Phenom II X6 1090T Black Edition

• Frequency: 3.2GHz
• Number of cores: 6 x physical
• Core: Thuban
• Packaging: Socket AM2+/AM3
• L1 cache: 64KB L1 data, 64KB L1 instruction per core
• L2 cache: 512KB per core
• L3 cache: 6MB accessible by all cores
• HTT: 200MHz
• Memory: Dual-channel DDR2/DDR3
• TDP: 125W

Recently I’ve come across a forum discussion where people were arguing about a PC configuration. It didn’t matter what particular PC it was about and for what purposes it was assembled, I just remember the argument in favor of the mainboard choice. The author of the forum post bought an Intel mainboard because Intel was a famous brand. This sounded like a lame argument to me. If a company makes excellent TV-sets, it does not mean that its digital cameras are going to be just as excellent.

Although there can be numerous points of view as to what processors are better, I guess nobody will question the fact that Intel produces good processors. It is simpler with chipsets since all makers of alternative chipsets for Intel’s modern desktop CPUs have abandoned that business, leaving Intel the monopolist in this field. And it must be noted that Intel chipsets were generally very competitive even when there still were competitors. Since Intel produces good CPUs and good chipsets, it seems logical to assume that their mainboard should be good, too. It can be some kind of an example to follow for the rest of the mainboard makers. However, Intel mainboards are actually often inferior to their opponents in functionality, setup options and usability. So, while choosing an Intel CPU and an Intel chipset is a natural decision, things are not so clear with Intel mainboards. The brand alone does not work here.

That’s why it wouldn’t be wise to blindly buy any Intel mainboard you see. You should first check out its capabilities and compare it with other products. And we, at X-bit labs, are ready to help you with test data for you to make your comparison and well-judged shopping choice. As you have already guessed, this review is about an Intel mainboard. It is called DH55TC. This microATX mainboard is based on the Intel H55 Express chipset and is designed for LGA1156 processors.

Gigabyte GA-X58A-UD7 Motherboard Review

Manufacturer: Gigabyte
UK Price (as Reviewed): £260.85 (inc. VAT)
US Price (as Reviewed): $349.99 (ex. Tax)

Like Asus and MSI, Gigabyte originally produced a huge range of X58 motherboards, but has since trimmed down the range to a few key models. The GA-X58A-UD7 is at the top of the range, both in terms of specification and price.

For starters, the GA-X58A-UD7 supports both SATA 6Gbps drives and USB 3 peripherals, so it’s as future-proof as it can possibly be. This isn’t enough for Gigabyte, though, as the GA-X58A-UD7 is crammed with other features. As well as two RAID-capable SATA 6Gbps ports powered by the Marvell 9128 controller, for example, there are three other SATA controllers powering a further eight RAID-capable SATA 3Gbps ports and two RAID-capable eSATA ports.

Fortunately, despite our initial concern, this doesn’t mean that the GA-X58A-UD7 takes an age to boot without RAID enabled on the controllers.

What’s more, although it has only two USB 3 ports, you’ll still find ten USB 2 ports and three FireWire ports dotted around the rear panel and PCB on headers. As it’s a full-sized ATX board, the GA-X58A-UD7 has six expansion slots – four 16x PCI-E, one 1x PCI-E and one PCI. Gigabyte claims that the first and third 16x PCI-E slots always run at 16x, but slots two and four will only run at 8x each.

However, we have doubts that the lanes are divvied up this way, as the X58 chipset has 36 rather than 48 lanes. Still, if it floats your boat, the GA-X58A-UD7 is certified for SLI, 3-way SLI and CrossFire.

The chipset, Southbridge and VRMs are cooled by separate heatsinks linked by a series of heatpipes. A key feature of the GA-X58A-UD7 is an optional waterblock that sits on top of the chipset heatsink, although there’s so little contact area between it and the heatsink beneath that we aren’t convinced that it will work very well. As such, it’s more of a gimmick than a genuinely useful inclusion.

Also included is an additional cluster of heatsinks and heatpipes that can be screwed onto the top of the chipset heatsink. However, it’s such a convoluted design that it barely performs any useful function. When we removed it, we found that the chipset temperature barely increased at all.

On a more positive note, the PCB layout is very good, with all the major components nicely separated. There are other pleasing touches too, such as all the SATA ports being mounted parallel with the PCB for neat cabling, a two-digit POST code display and six fan headers.

Dynatron Corporation was founded in Taiwan almost 12 years ago, back in 1991. Specializing on computer components cooling right from the start, Dynatron engineers invented the so-called MicroFin technology that according to them, is currently being used for production of majority of cooling systems. At this time, the company offers a wide range of coolers, fans, blowers, passive heatsinks, liquid-cooling systems and numerous accessories. However, we were particularly interested in only two solutions: EVO-11 and G950, which we are going to discuss today in our new review.

Dynatron EVO-11

The name of the new EVO-11 cooler comes from the word “Evolution” and clearly indicates that it is about some evolutionary development of a CPU cooling system. All sides of a relatively small cardboard box are covered with information, including a photo of the cooler, description of its key features, specifications, etc.:

Inside the box you find a plastic blister holding the cooler and accompanying accessories. The accessories bundle includes three retention kits for LGA775/1366 and Socket AM2(+)/AM3 platforms, retention screws and installation manual:

Dynatron EVO-11 really differs from most CPU coolers. But it is not its size (122 x 108 x 157 mm) or its weight (688 g) that are different, but its heatsink design. Let’s take a closer look at the cooler:

Well, it looks like a regular tower at first glance, but then you notice that this tower is shifted away from the central axis of the cooler. In fact, only the fan is hanging above the cooler base, while the entire heatsink is shifted forward:

Dynatron EVO-11 is designed with seven copper heatpipes 6 mm in diameter. The cooler heatsink consists of 53 aluminum plates pressed firmly against these heatpipes:

Each heatsink plate is 0.45 mm thick and they are spaced out at 1.8 mm from one another. The calculated effective heatsink surface is 6,110 cm², which is in fact pretty average for contemporary tower coolers.

One of the key peculiarities of Dynatron EVO-11 is the two-layer heatpipes structure in the base: four heatpipes in the lower layer and three in the top one:

Heatpipes lie in special grooves and are soldered to the base plate, as we can tell from the traces of soldering alloy on the edges. But it is not the most important observation here. It turns out that all three base plates are made of … aluminum! I have to admit that it is really hard to imagine, because thermal conductivity of aluminum is 1.88 times lower than thermal conductivity of copper and I couldn’t recall when we last tested a cooler with an aluminum base in our lab. But unfortunately, this fact is undeniable. Besides, take a look at how big of a gap there is between the top and the bottom layer of heatpipes. Considering comparatively low thermal conductivity of aluminum, it makes the top heatpipes layer more decorative than functional.