TechRadar UK Blogs: Performance PC

We remain unconvinced by Intel’s Atom. But it isn’t the only new chip vying for a piece of the burgeoning ultra-low cost PC (ULPC) market.

VIA has finally launched its much-anticipated Isaiah processor, which is now officially called the Nano. It’s looking very promising.

A WinChip winner at last?

Amongst all the acrimonious headlines revolving around Intel’s spat with AMD, it’s sometimes easy to forget that there is still another manufacturer making x86-compatible processors.

VIA’s CPUs haven’t been able to compete for the mainstream desktop market for some time – and indeed never really have. But they have been quietly ticking over in niche markets, particularly where low power consumption is most beneficial.

VIA purchased Centaur Technology from IDT in 1999, and with it VIA gained an x86 license. Centaur designed the WinChip, which was a Pentium-era competitor to Intel and AMD.

The WinChip was physically small, and therefore cheap to produce, but it never really won much attention. Its major failing was its floating-point performance. With entertainment software relying heavily on this – particularly video decoding and 3D games – its mainstream desktop appeal was limited.

VIA is no longer lagging behind

The WinChip evolved into the VIA C3 and C7, but it still had the same poor floating-point performance. Although the C3 started life as the Cyrix III, thanks to VIA also buying Cyrix off National Semiconductor, VIA switched to the Centaur design before launch, and has been developing this ever since.

The C7 regularly makes appearances in low-cost ultra-portables - the C7-M beats at the heart of HP's Mini-Note 2133, for example - where frugal power consumption and low cost are a boon. Just don’t go trying to play any games on one.

But the C7 doesn't handle Vista very well. This is where the Nano comes in, and why it is such a significant release for VIA. The Isaiah architecture has been promising vastly improved floating point performance ever since word got out about its development.

According to VIA’s own benchmarks of the Nano, it appears to be delivering on that promise.

VIA's Nano vs. Intel's Atom

VIA quotes floating-point performance tested using PassMark Performance Test 6.1. Pitting a 1.8GHz Nano against a 1.8GHz C7, the Nano gets 148.8 and the C7 just 73.8 – implying that the Nano is slightly more than twice as fast. The SiSoft Sandra XIIc (ALU Power Performance) result shows even more improvement – 887 versus 274, a three-fold increase.

All this bodes well. But the one thing VIA hasn’t given many details of is how the Nano measures up to Intel equivalents. The only comparison given is between a 1.6GHz Nano L2200 and a 1.6GHz Intel Celeron-M520. The Nano supposedly provides twice the performance per watt of the Celeron.

Intel’s Atom will be the main competition in the ULPC market, and that has so far proved to be rather puny. The 1.6GHz version is outperformed by a 900MHz Celeron-M in 1M SuperPI. So the Nano could be very competitive indeed with the Atom, at least in performance terms.

VIA also quotes some impressive scores for power consumption. A full range of five Nanos has been launched, from the 1GHz U2300 with a Thermal Design Power of 5W to the 1.8GHz L2100 with a TDP of 25W.  However, here Intel’s Atom may have an advantage. The Nano is quoted as consuming 100mW when idle, but the Atom supposedly needs only 30mW, and under 2.5W when fully utilised.

The ninny Nano revolution

With a current ceiling of 1.8GHz, the Nano isn’t likely to get close the performance of mainstream AMD and Intel desktop processors. It’s also currently still single-core, although VIA is planning dual-core versions in 2009.

But the Nano is VIA’s first 64-bit processor, despite the long time it is taking for the PC mainstream to move over from the current 32-bit norm.

So whilst this isn’t VIA’s answer to Intel's Core 2 Extreme and AMD's Phenom, the Nano is a much more competitive proposition than anything from VIA so far – particularly in the increasingly important low-cost portable market. It could also be the first sign of a three-horse race in processors again.

There have been rumours circulating for a couple of years about Nvidia releasing processors. I've touched on this before in the post - Nvidia inches closer to CPU business

AMD and ATI are now joined in holy matrimony. While Intel’s discrete Larrabee graphics solution is fast closing in. So Nvidia is facing the prospect that it will only have graphics cards and chipsets to offer, where its two main competitors will have the full kit and caboodle.

Both AMD and Intel are promising CPUs with integrated graphics for 2009, potentially cutting Nvidia out of a high-volume chunk of the market.

Nvidia's next generation graphics cards, the GeForce GTX 280 and 260, look likely to steal a march on ATI’s Radeon HD 4800 series. But that’s just at the high end. The true volume has historically been in integrated graphics, which is why Intel has the biggest percentage of revenue.

Where Nvidia has some pretty credible integrated graphics chipset options, graphics built into the CPU itself will make for an even cheaper proposition – particularly for business and notebook users. Just look at the popularity of Intel's GMA X3100, which appears in everything from Dell XPS laptops to the new MacBook Airs.

Unless Nvidia can offer an integrated CPU/GPU package in the future, it will end up only having competitive products in the discrete market.

Nvidia picks portables first

Nvidia’s first forays into the CPU business are not x86 desktop CPUs. According to slides leaked by The Inquirer from a recent editor’s day, Nvidia has two processor ranges approaching release, although one of them was actually already demonstrated at Mobile World Congress 2008, which took place in Barcelona back in February.

The Nvidia APX 2500, the one on show in Barcelona, is a full system on a chip based on ARM11, so combines CPU and GPU functions.

The APX 2500 is certainly flexible - it can allegedly encode and decode video at 720p and up to 14Mbits/sec. The ARM11 architecture also supports from one to four cores, and is aimed at everything from embedded car entertainment systems to PDAs and set-top boxes.

But Nvidia’s version was developed specifically for Windows Mobile, and supports Direct3D mobile and OpenGL ES 2.0. With GeForce-level graphics features, the APX 2500 is effectively a smartphone chip capable of HD video and 3D-accelerated gaming.

In contrast, the Tegra CSX 600 and 650 appear to be aimed squarely at Intel’s new Atom processor. This means that they are intended for use in portable devices in the larger UMPC, subnotebook and netbook categories.

Running at up 800MHz, the CSX will be capable of encoding and decoding video at 1080p, has support for hard disks and consumes less than 3W. But there's a problem. It won’t be able to run x86 code, such as Windows.

The x86 factor

At the moment, Nvidia can’t develop its own x86 processors anyway – it doesn’t have a license. There is an embedded 386SX in Nvidia’s product range called the M6117C, which integrates a ULi M1217B chipset. But if Nvidia wants to compete directly with Intel and AMD’s mainstream desktop CPU ranges it will need to obtain a full x86 license.

One analyst even suggested Nvidia might buy AMD to get an x86 license. But AMD’s deal with Intel means its own x86 license wouldn’t transfer, making this a rather pointless exercise.

More credible is the suggestion that Nvidia is interested in a deal with VIA, which already produces its own credible x86 clone thanks to the purchase of Centaur Technology from IDT in 1999.

A multi-core VIA Isaiah with Nvidia graphics integrated on die could be a pretty potent proposition, and a potential competitor for what AMD and Intel have planned in 2009. At the moment, though, VIA’s price is allegedly too high to make it happen.

Ever since Intel released the Core Microarchitecture in the middle of 2006, the processor business has been a bit like watching Mike Tyson go a few rounds with Rick Moranis – amusing, but rather a foregone conclusion.

At least the never-ending slugfest between ATI and Nvidia has been showing some signs of life. In June, however, it will be shaping up to be a prize fight of heavyweight proportions.

Over the last few months, more and more conclusive details have been trickling out about what ATI and Nvidia actually have planned and when.

It’s looking like ATI’s RV770 will arrive on June 16, with Nvidia’s reposte officially launched just a couple of days later. We’re back to the synchronised releases of old, and there's the potential for a real grudge match ahead in the second half of 2008.

Enter the ATI RV770

Back in February, NordicHardware leaked fairly comprehensive information about ATI’s RV700 series cards. According to the Scandinavian rumour-mongers, the flagship Radeon HD 4870 will have 480 stream processing units (SPUs), which is 50 per cent more than the current Radeon HD 3870’s 320 SPUs. So the new card could be at least 50 per cent faster than the old one.

This was little more than gossip a few months ago, but it has recently been corroborated by another leaked posting from a Folding@Home forum.

For those who aren’t familiar with this distributed computing initiative, developed by Stanford University, Folding@Home runs as a screensaver, processing biochemistry algorithms with spare clock cycles.

ATI has been working closely with Stanford University on a GPU-powered client, which was released in late 2006. So the Folding@Home team is likely to have similar inside information to other software development partners.

Engineering samples of RV770 PRO have been spotted in the wild running at 625MHz. But the 4870 could have a clock frequency of over 1GHz, a bit more than the Radeon HD 3870’s reference speed of 775MHz. ATI also looks likely to keep up the trend of memory early adoption, with GDDR5 on the cards for some versions of RV770.

The Nvidia GT200 strikes back

Nvidia has traditionally been better at keeping its GPU plans secret. For the last few months, the new GT200 looked set to be launched on June 18 as the GeForce 9900 series. This would have been another example of Nvidia's recently rather confusing naming strategy.

Fortunately, it now appears that Nvidia has realised the absurdity in having two cards called 8800 GTS with considerably different GPUs inside, whilst the 9800 was launched with a faster version of one of them. The current word on the street is that GT200 will be launched as the GTX 280 and GTX 260.

The actual specification of Nvidia's GT200 is a bit more mysterious than ATI’s RV770. But recent guesses have given it 240 SPUs, 32 raster units, 1GB of GDDR3 and a 512-bit memory controller. That’s nearly twice as many SPUs as the GeForce 9800 GTX.

Nvidia’s SPUs aren’t quite the same as ATI’s, and the fact that the GT200 may only have half as many as RV770 is no indication of relative performance.

Since G80, NVIDIA’s GPUs have run different parts of their core at different speeds, with the SPUs operating at 2.5 times the rest of the GPU. With the current 9800 GTX, the core is clocked at 675MHz but the shaders at 1688MHz. We haven’t seen any indication of GT200 GPU and memory speeds, though.

Let battle commence

ATI is also continuing with the dual-GPU strategy of its 3870 X2, although this is not quite what was rumoured for R700 last year, which had as many as eight GPUs on one card.

So in the red corner we have the ATI Radeon HD 4870 X2, with two 480 SPU-equipped GPUs running at over 1GHz each. In the green corner will be the Nvidia GeForce GTX 280, with 240 SPUs. Place your bets now. This battle is going to be a humdinger.

GeForce9900.png – The rumours are flying in the run-up to the two biggest graphics card launches of 2008

What was the first PC game which really grabbed your obsession? Counter Strike? Command & Conquer? Quake? In my case, I must sheepishly admit it goes back much farther than that.

Wolfenstein 3D was my first truly compulsive gaming experience – sneaky sessions on the office 486SX during lunch breaks. I’m not really that old – I started work when I was five. Honest.

It was obvious even then that PC gaming had huge potential – and huge potential to take over your life. The PC games industry has made huge leaps forward in the intervening decade and a half. It now points the way forward for entertainment in general, in particular the idea of paying monthly for an ever-evolving game experience.

The billion dollar question

Wolfenstein 3D was shareware. Its success propelled its creattors, id Software, into the game-developing big time, and the chance to sell bucket-loads of retail-boxed games. But in a recent report, the NPD Group calculates that the online subscription market for PC games has just topped $1 billion, surpassing retail PC games, which have been declining.

Only a few years ago, the online gaming business was worth a third of what it is now. Back in mid-2004, the Yankee Group estimated the annual value at $352 million, but predicted that online games would hit $1.1 billion by 2008 – which has proven to be absolutely spot on, looking at the NPD figures.

Where once it was sneaked FPS sessions during work breaks, now the majority of PC gaming income is derived from all-night sessions on a Massively Multiplayer Online Roll-Playing Game (MMORPG).

The top entry on the NPD list will surprise nobody - World of Warcraft. Try playing that during your work lunch break and you would soon be out of a job. And a girlfriend. And your ‘real life’ friends...

Apparently, there are 11 million monthly game subscribers. This actually isn’t that many when you consider there are now around 1.5 billion people worldwide using the Internet. So online interactive entertainment clearly has a huge growth potential ahead of it.

Being social makes all the difference

It became blindingly obvious when the first network-capable games arrived, like Doom II, that this was a very different gameplay to single-player.

Sure, you can discuss strategy with friends, but the post-game punditry of a good network deathmatch is on a different level. Then, as the first MMORPGs came out, such as Ultima Online, their arrival felt more like a networked version of very niche activities like Dungeons & Dragons.

But the social element of online play was also there, and in fact games like World of Warcraft and Lord of the Rings Online are proving very big business indeed – much bigger than their paper-and-funny-sided-dice predecessors.

Dungeons & Dragons, for example, is estimated to have made a $1 billion in its entire 30-year lifespan. Ironically (or maybe not), the fourth Edition of the RPG is due June 6, and will rely heavily on online resources such as D&D Insider.

The amazing thing is that PC gaming is not the biggest chunk of the gaming business – that is clearly consoles - GTA IV has already made $500 million on its own. Consoles are still undergoing the online hookup, but even more potential is available there. The Wii version of the BBC iPlayer is just the beginning. Wii Fit classes online, anyone? Maybe not, but you get the general idea.

Games like World of Warcraft have made the online subscription business bigger than retail PC gaming

AMD’s triple-core Phenom X3 is finally here. It may have sounded like an April Fool, but AMD has finally delivered on its announcement back in September 2007. Will the new holy trinity provide the renewed fortunes AMD needs to combat Intel?

The triple-core advantage

Although AMD’s Phenom X4 can’t quite compete with Intel’s Core 2 Quad even at the same clock speed, its real issue is the low frequencies currently available. The fastest Phenom X4 runs at 2.5GHz, whereas Intel’s Core 2 Extreme QX9770 is now on 3.2GHz. A clock speed issue is essentially a yield issue – AMD hasn’t perfected its Phenom production process sufficiently to produce enough high-clock CPUs just yet.

Although AMD would like to spin things another way, touting the benefits of three cores, the arrival of the Phenom X3 can only be an answer to this yield problem. AMD doesn’t have a special production line producing triple-core CPUs. They are all quad-core parts with one core disabled. The only reason AMD would want to do that is if one core isn’t performing up to the standard of the other three.

With its current designs, Intel can’t easily produce triple-core processors. Its quad-core parts are essentially two dual-core CPUs packaged together and joined at the FSB. So a triple-core design would mean two cores on one half, and one on the other – technically feasible, but hardly elegant.

So from AMD’s point of view, the Phenom X3 makes good sense. The Phenom X4 is hardly expensive – even the flagship 9850 Black Edition is only £150. But the entry-level Phenom X3 8450 is already below £100, putting it in the same league as AMD’s quickest dual-core Athlon 64 X2s.

Only a handful of Intel dual-core processors are cheaper – the Allendale 65nm ones with 2MB of L2 cache and 800MHz FSB rather than the usual 4MB and 1,066MHz of Conroe, plus the 45nm Core 2 Duo E7200, which has 3MB of L2 and a 1,066MHz FSB, rather than the usual 6MB and 1,333MHz respectively.

That’s likely to be an easy retail sell, now that true megahertz have been hidden behind mysterious model numbers in the processor market. For many punters, three cores for the same or less money than two sounds like a winner.

Intel strikes back

Intel’s response has been to slash its quad-core prices. And herein lies the rub for AMD. The 65nm Core 2 Q6600 can now be snapped up for a modest £134. The Q6600 has been a proven favourite amongst overclockers for nine months already.

So now the choice is dual-core, AMD triple-core for the same money… or add £40 and go quad-core. Intel’s other quad-cores are considerably more expensive, particularly the 45nm variety. The Q6600 is Intel’s single-handed spoiler.

If the Q6600 was available for £100, that would have been game, set and match to Intel. The current price leaves a little doubt, and a decision for potential buyers to make between price and an extra core.

So AMD should sell a few X3s. But only to those who don’t realise that an extra core is going to be of benefit primarily for tasks like professional 3D rendering. For most everyday software – and virtually all games – a higher-clocked dual-core processor will almost certainly be quicker.