VIA Cyrix III (Samuel 2) 600 and 667 MHz
Well, after you have read the title you might ask
us "why do you pay so much attention to the VIA company, and especially
to their not very successful Cyrix processors, which are not the
Cyrix at all, and even hardly work?" Actually, we are going to look
into it. Especially considering that there is the sound ground for
it - the release of the Samuel 2 processor which extends Samuel
core based VIA processors working at 533 and 550 MHz. The article
will start with words on the company VIA.
So, VIA, a prominent manufacturer and the strongest
competitor against Intel on the market of chipsets for mainboards,
set about promotion of the acquired processor work-outs and aims
at conquering up to 10% market of these processors. You might call
it nonsense. There is not only Intel, the whole Low-End sector of
the processor market is occupied by the cheapest and high-efficiency
solution from AMD with the Duron name! But in fact it's not that
bad. Right now VIA has bright perspectives since they are not burdened
with factories. The same factories that cost billions dollars and
which should be updated continuously according to the non-stop improvement
of technological processes. Besides, the clumsy giants Intel and
AMD have to produce on their the most modern factories only High-End
processors, and to invest the income into modernization of their
works. That's why in the nearest half a year the processors Celeron
and Duron wouldn't undergo any changes - they will be still based
on the 0.18 micron technology.
But there is such a monster of the Taiwanese semiconductor
industry TSMC (Taiwan Semiconductor Manufacturing Company) that
offers their services to VIA; their two factories work on the 0.15
micron technology.
It's exactly the latest technological process that
distinguishes the Samuel 2 based processors from the Samuel based
ones. Now the VIA processor is based on the most perspective technology
what allows tremendous reducing the die's size to 52 sq mm against
75 sq mm of the processor with the Samuel core and 90 sq mm of the
Celeron processors with a new core C0.
The 0.15 micron technology process makes possible
an easy increase of the clock speed of the processor, what VIA is
going to implement little by little, considerably reduced supply
voltage and energy consumption.
Let's start with the clock speed. The processor
VIA Cyrix III Samuel works from 500 to 667 MHz. Despite the quite
long 12-stage pipeline which in theory should allow overclocking
of the 0.18 micron processor up to 1 GHz and higher, the tested
processors poorly endured overclocking, and not because of the overheating
- the temp of the heatsink never exceeded 40 degrees C. The tested
550 MHz processor, intended for 100 MHz FSB, managed only 5,5 X
112 = 616 MHz, what hardly can be called satisfactory.
The Samuel 2 processors will work from 700 MHz
(according to the roadmap from VIA) and should quickly reach 900
MHz, and by the Q2 2001 we will see a new core EZRA, and the 0.13
micron process will have been polished - well, for VIA the tomorrow
seems quite rainbow. Today we can say only that VIA will easily
go through the increase of the clock speed for their processors
than the competitors in the Low End sector. In the nearest time
the Low-End processors from Intel and AMD might have uneasy life�
Of course not in performance, but in the clock speed; and this is
a huge step forward. Especially from the marketing point of view.
The second important issue is the power consumption.
At 1.5 V the 700 MHz processor with the Samuel 2 core takes only
5 W, what is less than even a mobile version of the processor Celeron
that works at 600 MHz.
The following diagram shows power consumption of
the processors:
For comparison - the usual (not mobile) 600 MHz
Celeron takes 17 W, while the 600 MHz Duron - 24,5 W.
The dissipated power constitutes around 70-80%
of the consumed one. Dissipated power shows how much heat must be
taken away by the cooling device. According to the figures we can
assume that the Samuel 2 won't need even a heatsink.
And if the VIA Cyrix III processors with the Samuel
worked without a heatsink, the Samuel 2 works wonders. Our processor
was operating during 24 hours without a heatsink and a fan. When
working in ordinary mode in business applications the core kept
the room temperature, with the business tests it warmed up a little,
and only with the Quake3 we felt that it really could get hot. The
processor reached 50 degrees C without any cooling, and no sooner
had we switched off the applications it got cold immediately.
And at last, the most interesting thing. The 52
sq mm core managed to house a L2 cache measuring 64 KBytes, which
should contribute the performance gain. Here you can see a table
supplied by the VIA company:
 |
Samuel |
Samuel2 |
Processor package |
CPGA
|
CPGA/PPGA
|
Socket |
Socket 370
|
Socket 370
|
FSB frequency |
100/133 MHz
|
100/133 MHz
|
Instruction support |
MMX, 3DNow!
|
MMX, 3DNow!
|
Clock speeds |
500~667 MHz
|
700 MHz
|
Technological process |
0,18
|
0,15
|
Supply voltage |
1.9/2.0 V
|
1.5 V
|
L1 Cache |
128 KBytes
|
128 KBytes
|
L2 Cache |
0 KBytes
|
64 KBytes
|
Pipeline stages |
12
|
12
|
We had two Samuel 2 processors in our lab, one
of them was a development type with the 667 MHz frequency, and the
other - a pre-production one working at 600 MHz. The core of the both
processors was marked C5B. You might ask why the frequencies start
not from 700 MHz? These processors were released in small lots,
specially for a preliminary estimation of the performance of the
future systems, and at the time of commercial shipping the frequency
will be exactly 700 MHz and higher.
The both processors are packed in a ceramic case
and the core is covered with a metallic lid which not only protect
the core but also serves as a heatsink.
The marking of the new processors is implemented
in the style of the Celeron. The Celeron processors with the same
clock speed, e.g. 300 MHz but differing in the presence/absence
of the L2-cache were marked as the 300A and the 300. Now the company
VIA marks the Samuel 2 with 600 MHz and 64 KBytes L2-cache as the
Cyrix III 600A.
The Samuel 2 will sell under the trade mark Cyrix
III, like its predecessor. But since this name hasn't justifies
the hopes, the large letters Cyrix now will be replaced by the large
VIA, and next to it you can see small letters Cyrix III.
The new processors Samuel 2 are promoted with the
help of Gigabyte. That's why for testing the new processor we have
taken the Gigabyte 6VX7-4X motherboard with the update version of
BIOS from Award which identifies the L2-cache.
Despite the fact that the processors are not serial
samples, they both are shipping with locked multipliers and are
intended for 133 MHz of the FSB. Quite scanty set of the GA-6VX7-4X
frequencies didn't allow estimating their overclocking possibilities,
though there is still something to say. The 600 MHz processor with
the multiplier equal to 4.5 managed to work at 150 MHz - 675 MHz
totally. The 667 MHz couldn't cope with 150 x 5 - 750 MHz - it refused
to pass the POST that's why we could reach only 140 x 5 = 700 MHz.
The overclockability is not impressive, so we hope that the series
models would show greater results.
Performance
The following testbed was used to estimate the
performance:
- Processors:
- VIA Cyrix III Samuel2 667A MHz, 133 MHz bus, Socket-370
- VIA Cyrix III Samuel2 600A MHz, 133 MHz bus, Socket-370
- VIA Cyrix III 533 MHz, 133 MHz bus, Socket-370
- Intel Celeron 366 MHz, 66 MHz bus, Socket-370 (for comparison
in performance was used as 550 MHz, 100 MHz bus).
- AMD Duron 600 MHz, 100 MHz bus, Socket-A
- Motherboards:
- Gigabyte 6VX7-4X on VIA Apollo Pro133A
- Epox 3VCA2+ on VIA Apollo Pro133A
- Soltek 75-KAV-X on VIA KT133A
- Memory: Hyundai PC133 128 MBytes
- HDD: Quantum Fireball CX 13 GBytes 5400 RPM
- CD-ROM: Panasonic 40x speed
- Video card: ASUS V7700 GeForce2 GTS (Core: 200 MHz; Mem: 166
MHz DDR)
Software:
- Windows ME final release build 3000
- NVIDIA Detonator 2 v6.34
- Ziff-Davis Winbench 99 v1.1 CPUMark
- Ziff-Davis Winbench 99 v1.1 FPU Winmark
- BapCo & Mad Onion SysMark 2000 Internet Content Creation
v1.0 patch 4B
- BapCo & Mad Onion SysMark 2000 Office Productivity v1.0
patch 4B
- idSoftware Quake III Arena v1.17 demo001.dm3
Well, the Samuel 2 performed rather well - the
higher clock speed and the additional L2-cache allow the processor
to run with confidence in business applications. The same L2-cache
allowed to brighten up the situation in game applications. The following
diagram shows a noticeable increase in performance of the 667 MHz
processor on the Gigabyte 6VX7-4X as compared with the motherboard
Epox 3VCA2+ brought in with the BIOS optimization for the L2-cache
of the new processor.
Nevertheless, the main disadvantage of the Cyrix
III with "A" letter is a slow co-processor.
The diagram above shows the results received while
overclocking of the Samuel 2 processors - but in this case the main
increase in performance was achieved mainly thanks to the bandwidth
of the bus "processor-memory".
That's why the strongest side of these VIA processors
is a usage of the 133 MHz mainboard. I hope that soon we will able
to see other architectural advantages, with the improved coprocessor
among them.
Besides, we should notice that the both processors
- 600 and 667 MHz - provide not just acceptable but even comfortable
game possibilities.
With the competent price policy, the Samuel 2 can
really gain its own market part - it has taken all the advantages
of its predecessor and has increased the performance.
High:
- low power consumption
- supersmall dissipated power in terms of desktops
- nonsusceptibility to mechanical effect
Lows:
- It's still hard to compete against the Celeron and the Duron
PS The author has ordered a Micro-ATX case with
the fan-lacking power supply unit for 135 W, and with the help of
the processor Samuel 2 he is going to assemble environmentally appropriate,
practically noiseless system case for comfortable writing of reviews
- in silence.