Power consumption, overclocking

Power consumption

We choose to check this point first by measuring the power consumption of the CPU block with a clamp meter via the ATX12V connector which is exclusively dedicated to the CPU. With our new method we precisely measure the power needed by the CPU power supply bloc, which has an efficiency comprised between 80 to 90%.


We measure the power consumption in four different situations with dual core processors: in advanced stand by, which means with energy saving options activated, in stand by without these options, in use at 50% with one session of Prime95 and then in maximum use with two sessions of Prime95.


First things first, compared to the E6400 we noted a logical reduction of power consumption in use. It is much lower than the A64 3800+, but we have to specify that an Energy Efficient 3800+ would have obtained much better results. Best results were achieved in stand by, and the L2 stepping shows there is potential. The reduction is quite significant and allows Intel to remain competitive to AMD in this domain.

Overclocking

Because of the FSB800, the E4300 needs a higher multiplying coefficient than the E6400 and E6300 to reach 1.80 GHz. It’s nine and similar to the E6600. This characteristic is really interesting in terms of overclocking since it is possible to reach high frequencies without using a high FSB, which are sometimes problematic for chipsets. The 975X, for example, is blocked around 400 MHz. This can also be problematic for memory with a chipset that can't be run out of synch like the P965 Express.

So in order to reach 3.2 GHz, required FSB will be 355 MHz instead of 457 MHz for the E6300. On paper, the E4300 is really excellent for overclocking, but now let's found out if it is really the case in practice. To do so, we have progressively increased the frequency and voltage of the processor with a P5B Deluxe. Each overclocking is validated by 15 minutes of two sessions of Prime95.

The voltage reported is the bios and in practice our P5B Deluxe is affected by a Vdrop of 0.05V approximately with a Core 2 Duo. It means that if we ask for 1.5V, the CPU in fact receives 1.45V in use.


With a starting frequency of 1.8 GHz, we have reached 3.06 GHz without changing the voltage. This represents 70% improvement! Then, we have increased the voltage progressively to reach 3,375 GHz (+87.5%), the increase required to reach 3.42 GHz was a bit more abrupt.


In short, this processor seems to be as comfortable for overclocking as the B2 and B3, even if we will have to wait for more test samples made on CPUs sold in shops to be absolutely certain of this information. Now let's find out if performances are equivalent!

Performances of the stepping, FSB800

Influence of the stepping

We have decided to measure the performances of the stepping L2 and compare them to the stepping B2 with a Core 2 Duo E6400 and a Core 2 Duo E4300 configured with a FSB1066 and a multiplying coefficient of 8. The speed of cache is studied via the RightMark Memory Analyzer, and the size of data written is indicated on the Y-axis:


Latencies are comparable between the two cores and graphics are practically similar. It is the same for reading, but it is slightly lower (-15%) for memory writing. This is quite strange and we can even wonder if it isn't due to the test card (P5W-DH Deluxe with bios 1707). The problem is that we no longer had the E6400 to check if the delta was still there with another card. Does it really have an influence in practice?


That doesn't have a great influence as you will be able notice. The performances between a E4300 o/c in E6400 and a real E6400 are pretty much similar.

Influence of the FSB800

The main difference between the E6300 and E4300 is the FSB, which decreases from 266 to 200 MHz. We felt that it was important to see the precise impact of this, and here are the results of the two CPUs side to side:


As E4300 is clocked 3.5% lower than the E6300 and performance gaps are quite close to this figure, we can believe that the FSB800 doesn’t have a great impact on the E4300’s performances in practice and despite the use of DDR2-800 4-4-4-12 for the two processors. In fact, the impact seems to be between 0 and 2% depending on the application.

The tests

We will now compare this processor with other dual cores with our usual test suite. Here are the configurations we used:

- ATI Radeon X1950 Pro / Catalyst 6.9
- 2 x 1024 MB DDR2-800 4-4-4
- 2 x Raptor 74 GB
- Windows XP SP2 French
- Socket 775 : ASUSTeK P5W DH (i975X)
- Socket AM2 : ASUSTeK M2N32-SLI Deluxe

In addition to the Core 2 Duo E4300, we integrated all core 2 Duos and several Athlon 64 X2s. Knowing that the E4300 is priced at $163, we will of course compare it to closer products in terms of pricing: E6300 ($183), Pentium D 945 ($163), X2 3800+ ($152) and 4000+ ($169).

3ds Max 9 and Maya 8

3ds Max 9 et Maya 8

For this test, we use two test scenes for Maya and 3dsmax developed by Yann Dupont of 3DVF (whom we thank) and using the MentalRay rendering engine. This choice wasn’t arbitrary since this engine is now available for both software and is most commonly used in production.

- The scene with 3dsmax is very heavy in terms of polygons and the number of objects. The objective was to test processor capacity and manipulate a heavy flow of data.

- Maya's scene is much lighter, but uses MentalRay's advanced lighting algorithms and employs the processors’ raw power in terms of mathematical calculations.


The E4300 provides very good performances compared to the competition whether it’s with 3ds 9 or Maya. With 3ds, the gap with the Pentium D 945 is rather impressive and the equivalent A64 X2, the 3800+ and 4000+, are much slower.

We noted the quite low performances of the 4000+ for this test compared to the 3800+. On the one hand the cache of the A64 X2 65nm is slower and the 4000+ configured with DDR2-800 can only run the memory at 350 MHz, instead of 400 MHz for the 3800+ because of AMD's management of memory (whole number divider based on the CPU frequency). So despite the frequency of 2.1 GHz, the 4000+ is slower than the 3800+ at 2 GHz.

With Maya, results are tighter: if the E4300 is the fastest the other processors aren’t that far away.

Mathematica 5.2 and WinRAR 3.61

Mathematica 5.2

For scientific calculations, we use the Wolfram Research’s Mathematica 5.2 and MathematicaMark2004 (benchmark integrated to the software).


You can take a picture, because it doesn't often happen. The Pentium D 945 at 3.4 GHz is much faster than the Core 2 E4300 at 1.8 GHz. Some calculations used in this test are very well adapted to the vectorization optimisations integrated to Mathematica 5.2 and work particularly well with Netburst architecture. The A64, however, doesn’t seem to be really at ease. The E4300 is even faster than the 5600+ in this test.

Tests integrated to MathematicaMark2004 were the following: Data Fitting, Digits of Pi, Discrete Fourier Transform, Egeinvalues of a Matrix, Elementary Functions, Gamma Function, Large Integer Multiplication, Matrix Exponential, Matrix Multiplication, Matrix Transpose, Numerical Integration, Polynomial Expension, Random Number Sort, Singular Value Decomposition, Solving a Linear System.

WinRAR 3.61

Since the 3.6, WinRAR includes multithread optimisations. We compressed in "best rate" a total of 588 MB of 493 Word and Excel files (69 MB), 22 Eudora e-mail box (251 MB) and one audio wav format (268 MB) file.


The A64 65nm aren't that comfortable in this test because of the relative slowness of cache and the under use of the DDR2-800 with the 4000+. The E4300 ends up being slightly faster than the Pentium D 945. The gap with the 3800+ and 4000+ is a bit more significant.

TMPGEnc 4.0 & DiVX 6.4

TMPGEnc 4.0 XPress

We are now using the fourth version of this MPEG-2 encoder. Compared to the third version, the latest integrates a couple of optimisations for the Core 2 and improves performances by approximately 5%. For this test, we continue to encode a 10 minute 16 second DV file to MPEG-2 format in 720x576 with an average bitrate of 4500 Kbits in two paths. The video preview display is activated during this test and the DV file is decoded via a Mainconcept codec, which is faster than the decoder in TMPGEnc.


This time, the Pentium D 945 is a bit faster than the E4300, but for both scenarios the results compared to AMD's offer are incomparable. Only the 5600+ provides comparable performances.

VirtualDub & DiVX 6.4

We use the 1.6.16 version of VirtualDub and the 6.4 version of DiVX. The latter improves performances by 8% with A64 X2, 10% with Pentium EE and 14% with Core 2. The same video is compressed as in TMPGEnc in Fast recompress mode, via DiVX 6.1 CODEC, in one path with an average bitrate of 1500 Kbits /s, b-frame and with best quality encoding performances. The video preview display is activated during this test.


Faster than the Pentium D 945, the E4300 is at the same level as the Athlon 64 X2 4800+.

Far Cry, Pacific Fighters, Flight Sim

Far Cry

Far Cry is still one of the heaviest FPS on the market because of the outdoor scenes in the training map, which we use.


Games have never been the strongest domain of Netburst architecture. The Core 2 E4300 isn't really bothered by the Pentium D 945. For AMD, results are quite dissimilar depending on the fabrication process. With the 65nm process, the 5000+ is faster than the E4300 whereas that with the 90 nm process, the 4600+ is fast enough.

Pacific Fighters

This time, the E4300 is comparable to the 4800+ 90nm or 5000+ 90/65nm and is much faster than is direct competitors. The Pentium D 945 is a bit slower than the 3800+ and 4000+.

Flight Simulator X

Here is the framerate obtained in a flight over New York after taking off from JFK. The level of details was "average high", as higher settings were really too slow.


Once more, the E4300 directly competes with Athlon 64 X2 having very high P-Ratings. It is right between the 4800+ and 5000+ 65nm. We noted that despite the quite low performances of the Netburst in this domain, the PD 945 provides higher performances than the A64 X2 3800+/4000+: the 3.4 GHz are quite helpful…

Conclusion

Conclusion

Since the release of the Core 2 last July, we never stopped repeating that the main default compared to AMD's dual core offer was the price. Indeed, because of the leader’s position, Intel hasn't changed the processor price list since this date and the entry ticket was at 180€ for the least expensive product, the E6300. Even if the manufacturer doesn’t reduce its prices, Intel today releases the E4300, a slightly less efficient model that reduces the entry ticket of approximately 20€.

Of course, the price gap might seem a bit low even if the performance gap with the E6300 is quite small. However, overall performances are quite honourable and superior to AMD's direct competitors and also from Intel's previous product lines, Pentium D processors at the end of their life. Also, this processor is part of an upgradeable platform and features very low power consumptions in stand by and in use.

If the E4300 is a good processor, it is mainly when pushed to its limits that it will be exceptional. The combination of a high multiplying coefficient and an architecture that can easily increase in frequency is a real ode to overclocking.

Of course, for now, we can't base our judgments on test samples. However, as we aren't the only ones to have successfully reached 3.4 GHz, we may believe that the CPUs in stores will be able like the standard Core 2 to reach frequencies between 3.2 and 3.4 GHz. Under these circumstances, would we ask for more? A lower price? Intel is planning on reducing it to $113 during the second quarter.

To counter this offer, AMD, unfortunately, doesn't have much to propose. Slower, using a less interesting platform for upgradeability and requiring high end memory to express their full potential, AM2 dual core processors do not provide better performances. In fact, the only advantage of AMD4s product line is to have cheaper mono-core processors. Intel will launch single processors using the Core architecture during the third quarter.

For dual core, AMD's only choice is to cut prices. The 4000+s compete with the E4300, but it should rather be the 4800+! But what would AMD do with slower products and what will the manufacturer do when the E4300 price will be reduced to $113? Maybe just pray for the release of the K8L?