Original Link: https://www.anandtech.com/show/914
VIA's P4X333 - If they only had a License
by Anand Lal Shimpi on May 18, 2002 6:05 PM EST- Posted in
- CPUs
There's no doubt that almost the entire Socket-A market is owned by VIA; the KT133, KT266, KT266A and KT333 lines have given VIA a larger percentage of market share than they've ever enjoyed making chipsets for other processors. You would think that with VIA's experience in building low-cost, high performance chipsets that they could easily clean up shop in making Pentium 4 chipsets as well, but things aren't always that simple.
VIA is still in litigation with Intel over whether or not they can legally produce any Pentium 4 chipsets, meanwhile they are happily selling the chipsets to any motherboard manufacturers that will buy them. Unfortunately, the major motherboard manufacturers are in great fear of the wrath of Intel and thus they won't touch any VIA Pentium 4 solutions with a 10 foot pole.
This was unfortunate for the P4X266 and P4X266A but as you're soon to find out, it's even more of a disappointment for the P4X333 and the forthcoming P4X600; the reason being that while the first two VIA P4 chipsets didn't really bring anything new to the table, the P4X333 and eventually the P4X600 will bring considerable value.
Currently the Pentium 4 chipset market is owned by Intel and what Intel doesn't have SiS takes with their 6xx offerings. All legal implications aside, it requires a truly impressive showing from VIA in order for them to even gain credibility in the Pentium 4 chipset market. It's time to find out whether the P4X333 is impressive enough…
AGP 8X and an Improved Memory Controller
The P4X333 North Bridge features two major improvements over the previous generation P4X266A solution. The first improvement is a DDR333 memory controller that is tuned for performance even beyond that found in the KT333. The P4X333's memory controller is the highest performing DDR333 memory controller of any currently available chipset. As you'll see by the performance numbers, it even gives PC1066 RDRAM a run for the money even while delivering just under 2/3 of the memory bandwidth.
Other than the improved memory controller, the P4X333 North Bridge features the first AGP 8X interface implemented in a mass production chipset. Before we get to the P4X333's AGP 8X compliance let's have a quick talk about the necessity of AGP 8X.
As we've explained in our recent graphics articles, before a GPU can start rendering pixels it must be fed raw data from the scene in the form of vertices. These vertices are passed on to the GPU over the AGP bus. In the past and even in current games, AGP performance isn't really a limitation since today's games make relatively limited use of polygons and are much more texture intensive. As more polygons are used in games, the sheer amount of vertex data being sent to the GPU over the AGP bus will increase and thus a higher bandwidth AGP bus may become necessary.
How much extra bandwidth does AGP 8X provide? The AGP 3.0 specification defines that a total of 8 transfers are made on every clock cycle with a frequency of 66MHz. Since the bus is still 32-bits wide this doubles the bandwidth of AGP 4X (1.06GB/s) to 2.1GB/s.
Unfortunately the reference board VIA sent out for testing would not even post with an AGP 8X card installed. We used the SiS Xabre 400 card and the board would not POST. We tried the same AGP 8X card in other AGP 4X boards and it worked fine. This may just be an implementation thing with VIA counting on not having to deal with it until AGP 8X cards start hitting the market in larger quantities. We got confirmation from SiS that the Xabre 400 works fine in forthcoming AMD and Intel AGP 8X chipsets.
The SiS Xabre 400 - The only AGP 8X card on the market
The P4X333 North Bridge interfaces to VIA's new South Bridge using a quad-pumped 8-bit V-Link interconnect. Now running at 133MHz and transferring data four times per clock, the interconnect offers 533MB/s of bandwidth between the North and South bridges. This increase in bandwidth isn't necessary and won't improve performance any but it won't hurt things either.
The First USB 2.0 South Bridge
Along with the new and improved North Bridge, VIA is hoping to lure the frightened motherboard manufacturers to their chipset using a new South Bridge. The VT8235 South Bridge is virtually identical to its predecessor, the 8233A with one major addition - USB 2.0 support.
The new South Bridge - We've shaded the V-Link interconnect in blue.
The integration of USB 2.0 into a VIA South Bridge makes a lot of sense but is a complete 180 from what VIA's position was at last year's Computex. During an interview with VIA's President, Wen-chi Chen, VIA took the position of supporting IEEE-1394 over USB 2.0. Luckily things have apparently changed and USB 2.0 has been integrated into the 8235 South Bridge. You can expect to find this South Bridge in all future VIA chipsets as well.
According to VIA, the USB 2.0 core was disabled in all reference boards that were sent out to the press. Citing driver issues for not enabling the functionality, VIA assured us that USB 2.0 is fully functional in all chipsets coming off the production line and the situation will be remedied immediately. Later this month Intel will release their ICH4 with USB 2.0 support, unfortunately ICH4 will only be paired with i845E and i845G chipsets at first.
Of course there is also an improved V-Link interface in the South Bridge to communicate with the new North Bridge at 533MB/s. Other than those two changes the South Bridge is identical to its predecessor, including its support for ATA/133.
What's in store for VIA South Bridges? The current goal is to integrate wireless networking technology onto the die of the South Bridge which would significantly enhance the value offered by the solution. However that is quite a while away considering that deciding on what standard to implement as well as getting the engineers to physically achieve that has yet to be accomplished.
The Reference Board & The Test
Our tests were run on VIA's P4X333 Reference Board; although the chipset is sampling to all of the major motherboard manufacturers, because of the aforementioned legal issues you won't see the chipset on any tier 1 boards. The biggest names you'll see that offer the P4X333 will be AOpen, Shuttle, and Soyo which isn't too bad but not great.
|
Windows
XP Professional Test Bed
|
|
|
Hardware
Configuration
|
|
| CPU |
Intel Pentium 4 2.4B
|
| Motherboard |
Gigabyte GA-8SRX - SiS 645 Chipset Gigabyte P4 Titan DDR - Intel 845 Chipset Intel D850EMV2 - Intel 850E Chipset VIA P4X333 Reference Board - VIA P4X333 Chipset |
| RAM |
1
x 256MB DDR333 CAS2 Corsair XMS3000 DIMM
|
| Sound |
None
|
| Hard Drive |
80GB
Maxtor D740X
|
| Video Cards (Drivers) |
NVIDIA GeForce4 Ti 4600 (28.32) |
Internet Content Creation & General Usage Performance
With this review we continue to use SYSMark 2002; SYSMark 2002 can be considered to be a much more memory bandwidth intensive version of the Winstone tests. The benchmark is split into two parts, Internet Content Creation which deals with content creation applications (Photoshop, Dreamweaver, etc...) and Office Productivity which is more general usage oriented (Word, Excel, Netscape, Anti-Virus, etc...).
The 2002 update changes things around a bit; first of all the benchmark's total scores are arrived at differently than in the 2001 benchmark. Windows Media Encoder no longer accounts for close to half of the Internet Content Creation test, rather only about 10%. There is also no need for a special Athlon XP SSE patch as the 2002 suite uses a version of the encoding dll that properly detects SSE support on all Palomino cores as well as Pentium 4 cores.
The rest of the benchmark is much more evenly distributed and it is much more memory bandwidth intensive than the old benchmark. The Internet Content Creation tests on average use about 600MB/s of bandwidth vs 300MB in SYSMark 2001. The Office Productivity tests are still stuck at around 580MB/s of memory bandwidth.
For more information on the tests and the applications used consult this whitepaper provided by BAPCo.
|
333
331
325
298Right off the bat we prove that the ICC portion of SYSMark 2002 isn't a Rambus-geared test as the P4X333 is able to provide more than enough memory bandwidth to offer performance that's virtually equal to Intel's 850E with PC1066 RDRAM. Keep in mind that DDR333 only offers 2.7GB/s of memory bandwidth vs. the 4.2GB/s of bandwidth offered by the 850E's dual channel PC1066 configuration.
Even with DDR266 the P4X333 is faster than both the Intel 845 and the SiS 645 chipsets.
|
181Although the P4X333 loses its top running spot in the graph above, the performance difference between it and the 850E is negligible (< 10%). You would be hard pressed to find any performance differences between the 850E and P4X333; the same cannot be said about the Intel 845 and SiS 645 chipsets which both bring up the rear and are at least 10% slower than the fastest P4X333 configuration.
3D Rendering Performance using SSE2
While 3D Studio MAX is SSE2 optimized, the level of optimization is nowhere near what NewTek reported with Lightwave upon releasing version 7.0b. The performance improvements offered by the new SSE2 optimized version were all above 20% using NewTek's supplied benchmarking scenes.
As we've shown countless times in the past, most 3D rendering applications require a good deal of memory bandwidth but not enough to illustrate differences between these platforms. Because of this we've stuck to only a single benchmark for the 3D rendering portion of our tests.
|
3D Gaming Performance
When it comes to most 3D games there's generally very little performance to be found by heavily optimizing for SSE2 or 3DNow! on either of these processors and thus the performance is mostly dependent on the overall platform (e.g. FPU capabilities, chipset, memory latency/bandwidth, cache latency/bandwidth, etc...). This makes gaming benchmarks the most important when doing a chipset comparison since it's much easier to see performance differences between chipsets.
We'll start off with our favorite 3D gaming benchmark - the Unreal Performance Test 2002. For an explanation of what this test is and why it is so significant, be sure to read our 15-way GPU Shootout that we used to introduce the test. In short, the benchmark uses the current build of the Unreal Engine (that will power games such as UnrealTournament 2003 and Unreal II) and serves as a great indication for future performance in games that use the engine.
|
The 850E takes the lead once again but by a very small margin (< 2%), illustrating the P4X333's very competitive demeanor. Once again the P4X333 outpaces everything slower than Intel's 850E.
|
With frame rates approaching 300fps at 1024x768 it's clear that Quake III Arena isn't video card limited any longer, making it the perfect platform test. The only platform here that really falls behind is the 845.
|
While the scores are much more reasonable under Jedi Knight 2 it's clear that the story told doesn't change at all. It's worth noting that just upgrading your chipset can help frame rates in JK2 more than upgrading from a GeForce3 Ti 200 to a GeForce4 Ti 4600.
|
Final Words
Provided that there are no issues with USB2 and AGP 8X, we can honestly say that VIA has produced the best DDR chipset for the Pentium 4. The performance of VIA's DDR333 memory controller is noticeably better than any other DDR solution on the market today, even VIA's own KT333. The unfortunate reality is that the P4X333 won't get the respect that it deserves with companies like ABIT, ASUS, MSI and Gigabyte unwilling to take a chance on facing Intel's wrath. All it would take is one major manufacturer to step forward and embrace the chipset and the rest would undoubtedly follow.
If the P4X333 isn't enough to break the ice of fear then VIA's P4X600, with its dual-channel memory controller may be what it takes. Although it's not clear how the costs of implementing dual 64-bit memory channels on a motherboard will change retail prices (we'd expect a noticeable increase) it would be very hard to resist a mainstream high-performance chipset that offered dual channel DDR support.
In the end, Intel has been quite effective in stopping VIA's entry into the Pentium 4 chipset market. With the P4X266/A it wasn't so much of a loss but with the P4X333 we're honestly being kept from a solid performing chipset with good features. Let's hope that solid board designs from companies like Shuttle and Soyo will keep the P4X333 from becoming an example of where engineering accomplishment doesn't result in market success.
Facebook
Twitter
RSS
