Tech Reviews, Guides, Articles for PC Hardware, Modding, and Gaming Enthusiasts

AMD XP 1700+ DLT3C JIUHB:0310XPMW

Posted on April 26th, 2003 | by admin

Introduction

This processor has me amped. The AMD XP 1700+ series is certainly not new, and it certainly
isn’t competing with the likes of Intel’s 3GHz HT processors, but it is proving
for some to be an overclocker’s dream. Not just any 1700+ will do though; a few
important changes have come about which will let us achieve a worthy overclock.
The most important change has been the move to the Thoroughbred core, instead of the
Palomino based cores of older XP series chips. The Thoroughbred core is on
a much smaller process than Palomino: 0.13 and 0.18 respectively. Furthering
the potential of this chip is the voltage rating – the DLT3C is rated at a stock core
voltage of 1.5v, as opposed to the more common 1.6v. These are all details we’ll tackle in a minute,
though.

When I originally ordered this processor, it was a toss up for me as to whether I should order
the XPMW or the WPMW. While I saw a number of very high overclocks on the XPMW chips,
the ‘general consensus’ was convinced that the WPMW is a more effective overclocker. I couldn’t
find anything to support this aside from rumor, so we are here today to find out!
I went ahead and ordered the JIUHB:0310XPMW from ExcaliberPC.com;
they sell OEM tray chips with guaranteed stepping, and much to my enjoyment the
proper CPU was shipped in great condition.

Let’s move on and see what this chip is really made of…

A package in my package

First of all, we will check out the detailed specs of this unit, and see how it looks up close.

Product: AMD Athlon XP 1700+ DLT3C JIUHB:0310XPMW
Application: Socket A
CPU Speed: 1.47GHz
Default VCore: 1.5V
Bus Speed: 266MHz
L1 Cache: 128KB (64×64)
L2 Cache: 256KB
Micron: 0.13

This chip was an OEM tray processor, as is evident by looking at the packaging.



Looking closer at the CPU you’ll see the stepping isn’t burned into the core as in the Palomino, but
instead labelled on the chip itself. Also, the top of this 1700+ is much more crowded than the
Palomino chips, a number of components were moved to the top.

Stepping on my core

In the picture it’s a little hard to make out, but the revisions/stepping
are as follows:

AXDA1700DLT3C
JIUHB 0310XPMW

0310 signifies the tenth week of 2003, so it looks like this chip was cut about a month
and a half ago (many thanks to leishi85 over at Icrontic
Forums
for correcting me on this code). Now I should also mention that I have heard
mixed impressions of these CPUs based on the XPWM and WPWM codes. I think the general concensus is
that the WPMW are better overclockers (note the W). But, looking again at
Lowyat.net, no significant
difference can be seen between the overclocks on these chips.

Let’s move on to our overclocking tests and see how this CPU can handle a little extra speed…

Overclocking

Let’s look at our test system first:

Motherboard: EPoX EP-8RDA+ A1/C1 Northbridge
RAM: 1×512MB Kingston HyperX PC3500
Video Card: MSI GeForce4 Ti-4600 @ 317.3/729.0
CPU Cooler: Thermalright SK7 & TT Smart Fan II
OS: Windows 2000

Since this isn’t a comparison review, I’m not concerned with how many FPS the
chip can pump out vs. this that and the other thing; I’m really only interested in what kind of
OC I can achieve and still run stable.

All tests are run at 200MHz FSB with the HyperX running in synchronous mode at DDR400. RAM timings
are at 2-4-4-8, 2.77VDIMM (except at 2.4GHz, I had to increase to 2.9 for a totally stable system).
These RAM timings are somewhat slow, but for the tests I would like to reduce the number of factors
involved in system stability.

As previously mentioned, this CPU comes at a default voltage of 1.5v, with a default core speed of
1.47GHz. Amazingly, I was able to bring the 1700+ to 2GHz solid on 1.5VCore. That’s 2,000MHz running
stock voltage! In other words, a full 36% OC without even upping the power to the chip. A
lot of people would be happy at 2GHz, but let’s move on and give it some more VCore. The following
chart shows our clock speed, voltage settings and idle temps at 2, 2.2, 2.3 and 2.4GHz:

Clock speed VCore Temp (c)
2GHz 1.5v 33°
2.2GHz 1.65v 39°
2.3GHz 1.7v 42°
2.4GHz 1.8v 45°

As you can see, temps rose appropriately as we increased VCore. The chip hit a wall at 2.5GHz,
refusing to boot all the way up to 1.9V. While it may boot at a slightly higher VCore, I am only
interested in stable overclocks. Idling at 45c is high enough, and another .1 VCore would certainly
bring temps near 50°C. This is too high for a stable overclock, in my opinion, therefore I’m stopping
at 1.8v and 2.4GHz.

We might guess that heat is already holding us back from 2,5GHz, but monitoring temps in BIOS they
were at 49c after a few failed attempts at booting. I have never seen a XP chip become unbootable
at anything less than 60°C. The Thermaltake Smart Fan II is working like a champ, running full blast
at 4800RPM and pumping out 75.7CFM of air. This HSF combo keeps load temps within a 5°C range of
idle temps in all situations I have monitored.

Regardless, at 2.4GHz we have achieved a nearly 1GHz OC! The 1700+ runs at
1.47GHz stock, which would work out to a 930MHz OC at 2.4GHz. This is a phenomenal overclock. There
is nothing to complain about here, and it just goes to show how much the T-Bred core helps this chip
out.

Now we can move on and see a few details of performance at each of these speeds.

Benchmarks

I am using the SiSoft Sandra Arithmetic, Multimedia, and Cache & Memory benchmarks to test the speed
of the chip at the various frequencies. Tests are run three times apiece, and the best score is
recorded. To test stability, 3DMark is run twice in succession, and instead of running even more
synthetic benchmarks I put it to the real test – two hours of Battlefield 1942! If the system
depicts any problem, quirks, or unacceptable latencies, the score is discounted and I move on.

Arithmetic Benchmarks

Let’s look at our results then, starting with the Arithmetic test. The Sandra Arithmetic tests
stress the ALU (Arithmetic Logic Unit) and FPU (Floating-Point Unit) of the CPU.

SiSoft Sandra Arithmetic Benchmark
Clock speed ALU FPU
2GHz 7508 3010
2.2GHz 8279 3311
2.3GHz 8580 3461
2.4GHz 9032 3613

Multimedia Benchmarks

Next are the Multimedia tests, which stress similar aspects as the Arithmetic tests yet place more
stress on the MMX and SSE/SSE2 capabilities of the processor.

SiSoft Sandra Multimedia Benchmark
Clock speed Integer Float
2GHz 11093 11848
2.2GHz 12196 13006
2.3GHz 12745 13624
2.4GHz 13305 14234

Cache & Memory Benchmarks

And finally, the Cache & Memory test results. The cache and memory benchmark is testing the speed
of the CPU cache, chipset and memory subsystem. While this is a useful indicator, you’ll notice a
strange discrepancy in the figures. As soon as we hit 512kB transactions, the lower clocks start
showing faster results! Whether this is due to some kind of latency introduced with the huge
transactions at higher speeds, or if I personally made a mistake in my testing, I’m not sure. I
intend to re-run all of the Cache & Memory tests, paying especially close attention to ensure that
these numbers are correct. There is always the possibility that a BIOS setting was inadvertantly
changed, or otherwise. Rest assured I will do my best to retry these tests and update in the case
of a different set of scores.

SiSoft Sandra Cache & Memory Benchmark
Transaction Size 2GHz 2.2GHz 2.3GHz 2.4GHz
2kB 16235 17945 18712 19597
4kB 17307 19044 19911 20782
8kB 17787 19578 20472 21363
16kB 14972 16480 17233 17996
32kB 12429 13868 14306 14940
64kB 11612 12777 13345 13943
128kB 7635 8398 8783 9162
256kB 6288 6916 7236 7550
512kB 1305 1295 1423 1361
1MB 1217 1169 1348 1279
4MB 1226 1142 1318 1256
16MB 1290 1098 1261 1178
64MB 1253 1085 1256 1180
256MB 1228 1075 1242 1168

Click on this chart to see an expanded version!

Aside from the strange difference at 512kB transactions and above, results are as
we would expect. We can pick out the sweet spot for this chip, too; at 2.4GHz and 8kB transactions
we are seeing an extraordinary transfer rate of 21,363MB/sec!!

Those are all of the tests I’m reporting on today. It’s fairly obvious that 2.4GHz is faster than
2GHz, but the point of this was to see the OC’ability of this chip, and also to compare it at a few
levels to see how things like voltage, latencies, and other factors can affect performance as you
continue to up the OC.

Conclusion

What can I say, I think it’s pretty obvious at this point the 1700+ DLT3C JIUHB:0310XPMW is an
insane overclocker! I think these numbers help put to rest the XPMW vs. WPMW controversy, too; or
atleast show that if any difference does exist, it is most likely a very small one.

Compare this processor to the XP chips based on the Palomino core, and you will see that AMD did a
very good thing moving to the Thoroughbred. My previous chip, an XP 1900+ rated at 1.6GHz,
would do a maximum of 1.8GHz before it choked. While problems were recorded with the first
round of Thoroughbreds, the newest revision “B” cores do not suffer from this problem. We might also
want to take notice that there are two different “B” cores available, the “low B”, as we have seen
here in the XP 1700+ processor, and the “high B” (or just “B”) available in XP 2700+ chips and
above.

I’ll leave you today with a few screenies of WCPUID and USDM, so you can see that the chip is
actually running at my stated speeds. Also included is a Sandra pure Memory bench, just for kicks.
Fare well tweakers, and until we meet again, may the hertz be with you!



Back to Fury-Tech.com

Add to Del.cio.us RSS Feed Add to Technorati Favorites Stumble It! Digg It!
    www.sajithmr.com

Best Places to Buy AMD XP
Store Rating Store Name Price Visit Store

214 Store Reviews

Target
$ 999.99 Go!

2155 Store Reviews

eCOST.com
$ 1099.99 Go!

Related posts:

  1. AMD XP 1700+ DLT3C JIUHB:0310XPMW Overclocking
  2. AMD XP 1900 & Thermalright SK 7
  3. Digit-Life Review: Intel Pentium XE 955: Ultima Ratio of the Kings
  4. Vantec iCEBERQ4 VGA Cooling Kit
  5. Kingston HyperX PC3500 DDR SDRAM (KHX3500)

Tags: Case, CPU, mod, motherboard, Ram

Leave a Reply