Intel Pentium 4 2.8 GHz. At full speed towards 3 GHz
Foreword
The idea of the last Pentium 4 processor is back now again over three months. Three months in which Intel almost all benchmarks over the competition could – dominate games as well as applications felt extremely comfortable with 2,53GHz on the Pentium 4. This was no surprise, had increased but finally with the idea of this processor of also the front side to 533 MHz Bus (FSB), which is responsible for the communication between the processor and memory controller (Northbridge), Intel, which could increase the bandwidth of this link of 3.2 GB/s on 4.2 GB/s. With new Rambus memory which now effectively is clocked (PC1066/PC4200) with 533 MHz and so now in turn is connected with 4.2 GB / s on the Northbidge, perfect harmony between the Pentium 4 and Rambus, which resulted in significant performance jumps, was how we bus of 533 MHz also could prove it in our test of 2,53GHz Pentium 4 with a front side.
The biggest competitor of Intel, which must be presented here certainly not continue countered recently with two new models: the Athlon XP 2400 + as well as its faster journeymen, the Athlon XP 2600 +. Not only that AMD with these two processors in our opinion indirectly acknowledged, that the model number counter the official position is looking just comparison to Intel’s Pentium 4. No, also the speed Crown could conquer so back AMD is in some fields. However, and that we want to take just beforehand, couldn’t AMD no samples for this test us to provide and you had also a XP2200 + (‘Thoroughbred’) seemingly not in stock, so had here a 2100 + to 2200 + “gelockter” to serve ‘Palomino’. Because both processors but only in the die size and production structure, the results still are those of Thoroughbreds.
With the idea of 2,8GHz of fast new Pentium 4 on Monday, 26 August, Intel would like to ensure clarity on the fronts. But the clock increased by 10.5 percent is sufficient to assign AMD in its place and can benefit from applications at all of the tick increased so high? We will resolve these and other questions in this article.
The new
With the introduction of FSB533 the chip giant Intel was present three processors, which benefited from the higher bandwidth. In addition to the 2.53 GHz clocked top model of the Pentium was presented at that time 4 2,4B and a 2.26 GHz fast model. Since the 2.4 GHz Pentium 4 with a FSB400 is, is to identify the “B” in the name of the FSB533. Intel pronged thus currently so more or less. Although you would like to push the old FSB400 from the market as quickly as possible, can the owners of older Pentium 4 motherboards which support officially only this front side bus, not simply reject.
In addition to the Pentium 4 with 2.8 GHz (FSB533) and the equally new 2.66 GHz model introduces two new computing servants Intel therefore together with these processors, running on the old front side bus. This would be one of the 2.5 GHz chip and the other a 2.6 GHz Pentium 4. Thus the chip giant has from the sunny San Jose advanced its product portfolio to four processors.
- 2.80 GHz – 21 x 133 MHz (FSB533)
- 2.66 GHz – 20 x 133 MHz (FSB533)
- 2.60 GHz – 26 x 100 MHz (FSB400)
- 2.50 GHz – 25 x 100 MHz (FSB400)
The fastest model can be run only on current motherboards which offer a front side bus of 533 MHz. However the possibility here that Intel with the idea of presenting next faster processor, a Pentium 4 with 2.8 GHz, which can be operated with FSB400. This is not clear but this from us roadmaps of the processor’s manufacturer. Moreover, addition here make matters worse, that already the now featured new FSB400 can be awakened models not easily on current as well as slightly older motherboards to life. But this later more.
New stepping
With the launch of new processors, Intel has raised also the CPUID to three version numbers from 4 to 7. Thus, the new processors are based on a revised processor stepping, which is obviously essential for higher clock rates. But what is this at all, a “processor stepping”? As the name already suggests, describes the stepping (step – available: level) more or less the State of a processor. Because it is dedicated to such highly complex components such as a processor not from small errors, often known bugs will be ironed out through a revision of the processor core. This stepping must not only on the core, the so called ‘the’, refer. Also changes in the package can result in a new ‘processor ‘level. Intel identifies this Steppings General letter followed by a number, to distinguish them clearly.
In general it is in A0 stepping the first version of a processor that will pass through during development. Rarely, the main processor unit from the beginning is so perfect that she can just go into mass production. At the 0, 13µm of produced Pentium 4, codenamed Northwood, was in the B0 stepping ripe for production. This stepping is often also called nB0, so that it does not cause confusion with the in 0, 18µm of manufactured Willamette core of p4s comes. Generally, you can see a new stepping of Northwoods by a leading small “n”. So, for example, a small t”" is used in 0, 13µm manufactured Tualatin core (Pentium III) to distinguish from the previous Coppermine core which.
On the classical nB0 stepping we are answered already in our previous article. Since then, Intel has made almost twice newly launch this processor. But the chip giant has changed only once the CPUID with the new processors. But a slightly revised core was used in 2,4B and 2,53GHz Pentium 4. The changes were already at that time noticed us during our test, information, available to us but only now available. What had changed?
Reminder: on the capacitors on the bottom of 2,4B tested by us and 2.53 GHz Pentium 4 were changes in the joints of the capacitors visible, differed slightly from the previous Pentium 4 processors. The processor CPUID remained unchanged.
Recently, more precisely said on 22 July, Intel published a document that could clarify our issues. There are now two versions, a nB0 non-shrink and shrink nB0 with the nB0 stepping. Of all processors between 1.80 and 2.53 GHz both have to and from the 2.6 and 2.66 GHz Pentium4 only the revised model available. However, what was here shrunk (shrink – Filipino: shrink) and how was this suddenly possible? All in all the core has become slightly smaller. Of initial 146 mm square, he could be reduced by a tighter placement of the library facilities and the removal of ‘White space’ by 10 percent to 131 mm². A new arrangement of the functional units of the processor was not needed. As a result had to also the capacitors to be moved slightly, their number and capacity remained unchanged. The electrical properties of the processor, such as, for example, EMC, have not changed however. Also the 23 known errors (errata), which brought the processor itself, remained unchanged in their number. In particular, Intel has probably seen not the need for a new stepping ID.
New stepping (continued)
When the 2,8GHz Pentium4, the thing looks however differently. In the heart of the processor works for the first time, namely a processor core that is adjusted to more errors. Information on what Intel exactly could eliminate errors, no occurred but us up to the deadline. The fact is that there are less than 23 only. As a result Intel were worth these changes now a new stepping. The CPUID was increased from 0F24h to 0F27h (stepping from 4 to 7). The new stepping the name nC1 and also comes along with other changes. On the one hand the number of capacitors, which provide the transistors with voltage and compensate for fluctuations in supply was reduced from 15 to 12. Since the new capacitors are slightly larger, it can be assumed that their capacity was increased slightly to can as many charges as before save in total. Presumably, Intel this saves a few cents in the production of the processor.
Also the operating voltage of the processor was not not affect Intel. All Pentium4 processors of nC1 Steppings be addressed with a VCore of 1,525 volts. Thus it increased slightly the tension around 0.025 volts, so sufficient stability at higher clock rates to ensure, as the signals through the higher voltage be ‘clean’. You will also to similar steps when one processor on its specifications operate (overclock) would like to. Parallels are pushing up here already. Because it is frequently misrepresented, we would point out expressly that the Pentium 4 in operation is almost never supplied with this voltage. With increasing CPU usage she also reduces input voltage to the processor. We have explained this principle already closer in another article. At this point I would like to point out for more information.
Also the translation lookaside buffer (TLB) remained at the new stepping not untouched. The instruction TLB can contain now 128 instead so far 64 entries. In the TLB are not even stored the data, but, simply put, the way to them or containing the “maps”; in English, so too often maps and data/statements used in the recent past are kept in the TLB. Somewhat more specifically, these maps form the link between the (known to the processor, issued by the program) logical memory address of the data, or the physical memory addresses (required by the processor). In the TLB, the physical addresses the most frequently used data to be used so constantly.
Update: is the information according to the us just a reporting issue. Always 128 entries in the TLB had can accommodate the Intel Pentium 4.
All in all offers the nC1 stepping thus the guarantee for even higher clock rates. In addition to the Pentium 4 with 2,8GHz, which is available only with the nC1 stepping, all other Pentium 4 processors between 1,8A GHz and 2,66GHz in the nC1 stepping available will be available. When you buy a Pentium4s you should make sure therefore to get one of these models. On the heat-spreader (correct IHS) this can be through an S-spec five-digit code identify exactly. Also, a distinction by the larger capacitors on the bottom is possible. The following table to bring some light into the darkness of the S-spec codes.
< div >
| Processor | nB0 non-shrink | nB0 shrink | nC1 |
|---|---|---|---|
| 1.80A | SL62P | SL66Q | SL6LA |
| 2 2A | SL5YR | SL66R | SL6GQ |
| 2.20 | SL5YS | SL66S | SL6GR |
| 2.26 | SL67Y | SL6D6 | SL6DU |
| 2.40 | SL65R | SL66T | SL6GS |
| 2.40B | SL67Z | SL6D7 | SL6DV |
| 2.53 | SL682 | SL6D8 | SL6DW |
One downside to tiny also has the new stepping: the thermal design power of the new models is above those of its predecessor, since the operating voltage had to be raised.
New production
Already in our article on the Pentium4 2,2GHz we had reported that Intel will convert after 200 mm wafer 300 mm wafer manufacturing.
Wafers are round, polished Silicon slices and form the basis for the production of the chip. By moving to larger wafer manufacturing costs are reduced, because a lot more this on a wafer place find and on the other hand the blending at the edges, which this logically occurs in a round silicon wafer and square percentage considered on a larger wafer less fails. So, 300 mm wafers offer a larger by a factor of 2.25 area than 200 mm wafers. The cost for a single wafer grow but not by this factor, so that here already first cost be saved.
The new wafer moved himself but also other changes in the production. The 200 mm wafers by hand could be transported, this is no longer allowed for ergonomic reasons for 300 mm wafers. The enlargement of the area by a factor of 2.25 draws also weight increased by a factor of 2.25 in.
An automatic and mechanical assistance was therefore required to deal with the 300 mm wafers. This can be partially naturally short delays in the production process, because simple handles with mechanical support now only can be executed. In addition, the whole factories had to be adapted course at the new wafer and the conversion of complete production parts was inevitable.
Such switching course only makes sense if the production with 300 mm wafers guarantees identical performance and quality at better yield. At the beginning of the development, the yield with 300 mm silicon wafers was even lower than with 200 mm wafers. In the course of the further improvement and adaptation of production technology, the 300 mm wafers in this area included but quickly to their predecessors.
But why is this change in production at the new Pentium4 not uninteresting models? Already at that time for the conversion of the production process of 0 18µm on 0 13µm this could be made more per wafer. As also the wafers were converted from 200 mm to 300 mm, this made double impact. 144 Still fit on a 200 mm wafer this the old Pentium4 – Willamette, so already 200 could be accommodated on a 300 mm wafer this this. A Willamette-which was still 217 mm m² big. Through the new 0 13µm, technology reduce the DIE size of Northwoods despite double L2 cache on 146 mm². So came so far 484 Northwood-under this on a 300 mm wafer.
But as we already reported in the section the new stepping, Intel has reduced this surface by a tighter placement of the library facilities and the removal of “Empty land” by 10 percent to 131 mm². Thus, the yield of this per wafer was once again increased and reduced the cost of production once again.
Overview
Northwood or Athlon XP – if it would be so easy. However, under the simple name, a number of different core types and Steppings currently stuck. The following table shows on the most important details of all currently available processor types.
< div >
| Features | Pentium 4 (FSB400) | Pentium 4 (FSB533) | Athlon XP | Athlon XP |
|---|---|---|---|---|
| Core | Northwood | Northwood | Palomino | Thoroughbred |
| Production | 0, 13µm | 0, 13µm | 0, 18µm | 0, 13µm |
| Socket | Socket 478 | Socket 478 | SockelA | SockelA |
| Clock rate (stepping) | 1600 MHz 1800 MHz 2000 MHz 2200 MHz 2400 MHz 2500 MHz (nC1) 2600 MHz (nC1) |
2266 MHz 2400 MHz 2533 MHz 2666 MHz (nC1) 2800 MHz (nC1) |
1333 MHz (1500 +) 1400 MHz (1600 +) 1466 MHz (1700 +) 1533 MHz (1800 +) 1600 MHz (1900 +) At MHz (2000 +) 1733 MHz (2100 +) |
1800 MHz (2200 +) 2000 MHz (2400 +) *. 2133 MHz (2600 +) *. * Thoroughbred “B” |
| Transistors | 55 Million | 55 Million | 37.5 Million | 37.5 Million |
| DIE size | 146 mm2 131 mm² (nC1) |
146 mm2 131 mm² (nC1) |
128 mm2 | 80 mm2 (T-bred “A”) 84 mm2 (T-bred “B”) |
| Front side bus | 400 MHz QDR | 533 MHz QDR | 266 MHz DDR | 266 MHz DDR |
| L1-execution cache | 12,000 µ OPS | 12,000 µ OPS | 64 KB | 64 KB |
| L1 data cache | 8 KB | 8 KB | 64 KB | 64 KB |
| L1-stroke | CPU clock | CPU clock | CPU clock | CPU clock |
| L2-cache | 512KB | 512 KB | 256KB | 256KB |
| L2 connection | 256-Bit | 256-Bit | 64-Bit | 64-Bit |
| L2-cache clock | CPU clock | CPU clock | CPU clock | CPU clock |
| Hardware data prefetching | Yes | Yes | Yes | Yes |
| VCore | 1.5 Volt 1,525 Volt (nC1) |
1.5 Volt 1,525 Volt (nC1) |
1.75 Volts | 1.65 Volts |
| Instruction sets | MMX SSE / SSE2 |
MMX SSE / SSE2 |
MMX / 3DNow! 3DNow! + / SSE |
MMX / 3DNow! 3DNow! + / SSE |
| Temperature diode | Yes | Yes | Yes | Yes |
| Multiprocessor capable | No | No | No | No |
| CPU architecture | 20-stage pipeline | 20-stage pipeline | 15-stage (FPU) 10-stage (ALU) Pipeline |
15-stage (FPU) 10-stage (ALU) Pipeline |
What are the news of our last article now? On the part of Intel’s only the introduction of new Steppings ‘nC1′ reveals in the end which which we have already dealt with. AMD has the often maligned Athlon XP Kern Thoroughbred’ duly revised and placed in the “B” version. The power consumption could be so significantly reduced, increased the clock rate. The area of the rose but by good 4 mm². Unfortunately, CPUs were US based on this core, as already mentioned, for a test not be available.
The properties of the Intel NetBurst technology we have also covered in old articles to meet and refer again to the six tags of this architecture that makes possible such high Takraten.
- Hyper pipelined technology
- Rapid execution engine
- Advanced dynamic execution
- Execution trace cache
- Quadspeed system bus
- Streaming SIMD Extensions 2 (SSE2)
Power consumption
The power consumption of processors has increased in the past months and years. By finer production structures and a resulting, lower operating voltage, one could turn away at least a processor with a power consumption of 100 Watts in the desktop area. However, you will slowly but surely also this brand something closer. The manufacturer of modern cooling solutions on these increased requirements can set, it is common to specify a value for each processor, which describes the required cooling power. Here the term has become thermal design power (TDP). This value specifies the usual heat dissipation of the processor on the also the cooling solutions need to be created.
In determining this TDP value, there are however large differences between Intel and AMD. In the technical documents, Intel remains silent out entirely on the formation of the value – AMD is exceptionally something open. This value not readily can be derived from the maximum values specified in the technical documents for power consumption and maximum temperature.
At this point, we should take a look at the usage of the Pentium4 and Athlon processors. This 4 Willamette, Northwood nB0 and Northwood nC1 must be between Pentium differentiated. As already mentioned the nC1 stepping of Northwood at same time consumes more power, because the voltage from 1.5 to 1,525 is increased.
Power consumption
Information in Watts (W)
|
Even if the maximum power consumption is rising continuously, so the current Pentium 4 with 2,8GHz consumed 18µm Willamette core still less than the last Pentium 4 with 0. Thus also the most mainboards can cover processor without problems of 2,8GHz hunger, if they could run even a 2,0GHz Willamette stable.
Heat protection
So that the CPU just expensive acquired immediately vanishes the customer not after a fan breakdown in a puff of smoke, Intel Pentium 4 has powered an effective protection. Differently than was the case when AMD the case, this need not the help of the mainboard (keyword: ASUS C.O.P). The protection mechanism is made here of two components.
The first protection mechanism of the Pentium 4 is the so-called ‘ thermal monitor’. It is composed of a temperature sensor directly in the core, a signal (PROCHOT#), indicating if the processor has reached its maximum temperature, and a control mechanism (thermal control circuit TCC) which can affect the CPU temperature through the adjustment of the CPU cycles. Once the CPU therefore has reached a set fixed temperature, the PROCHOT# signal is activated, TCC deploys and omits some bars of the CPU. Around every second measure is usually omitted. This is of course to keep in mind that this has a direct impact on the performance. Is every second beat is omitted, to the performance be reduced. Much more important in this context also the CPU temperature decreases significantly, and the processor is protected from corruption will however. The fixed temperature value is again below, consists of the PROCHOT# signal and the processor will work again with full power. From which value uses the so-called Throttlen, is precisely tailored to each individual CPU.
Help should however this mechanism no longer out on hot days, at a constant load or a cooling failure, then the emergency switch-off attacks.
The second temperature sensor is intended only for absolute emergency if the cooling fails and omitting several bars brings no success. As soon as the Silicon reaches a temperature of about 80-90 ° C, reacts to the Pentium 4 with the THERMTRIP# system bus signal and turns itself completely off. Because THERMTRIP# operates completely independently of the processor, it produces no bars. The signal remains active as long as until the engine taken from the power source and has been a complete reset.
We have tested this function on a 2,0GHz, 2, 4 GHz (FSB533) and the 2,8GHz processor by we have completely dismantled the cooler short hand. Each after about 5-15 seconds enacted the emergency shut-down, and turned off the CPU.
< div >
| Processor | Emergency shut-down at |
|---|---|
| Pentium 4 2,0GHz | 83 ° C |
| Pentium 4 2.4 GHz | 85 ° C |
| Pentium 4 2,8GHz | 90 ° C |
Here too, the decisive temperature differs from CPU to CPU. Also permanently this function to prevent the processor from damage seems even after a good dozen emergency shutdowns is still as reliable as the model with 2,0GHz as ever and ever is running.
Platforms
The newly introduced processors make necessary although no new chipsets, however, would we give you a small overview of the chipsets, which officially come for the new processors in question:
< div >
| Features | Intel 850E | i845E | i845G | |
|---|---|---|---|---|
| Supported CPU sockets | Socket 478 | Socket 478 | Socket 478 | |
| Northbridge features | ||||
| Northbridge | KC82850E | KC82845E | KC82845G | |
| Multiprocessor support | No | No | No | |
| Front side bus | ||||
| 400 MHz | Yes | Yes | Yes | |
| 533 MHz | Yes | Yes | Yes | |
| Memory clock | ||||
| 100 MHz | Yes (400) | Yes | Yes | |
| 133 MHz | Yes (533) | Yes | Yes | |
| 166 MHz | No | No | Yes (unofficial) | |
| 200 MHz | No | No | No | |
| Asynchronous memory clock | Yes | Yes | Yes | |
| Memory slots (max) | 4 | 3 | 2 | |
| Storage support | ||||
| SDRAM | No | No | Yes | |
| DDR-SDRAM | No | Yes | Yes | |
| (Dual)-Rambus | Yes | No | No | |
| AGP support | ||||
| 1 x | No | No | No | |
| 2 x | Yes | Yes | Yes | |
| 4 x | Yes | Yes | Yes | |
| 8 x | No | No | No | |
| Southbridge features | ||||
| Southbridge | Intel 82801 BA | Intel 82801 DB | Intel 82801 DB | |
| Disk controller | ||||
| ATA 33 / 66 | Yes | Yes | Yes | |
| ATA 100 | Yes | Yes | Yes | |
| ATA 133 | No | No | No | |
| Number of PCI-slots (max) | 6 | 6 | 6 | |
| USB ports (max) | 4 | 6 | 6 | |
| USB 2.0 | No | Yes | Yes | |
| FireWire | No | No | No | |
| Other | ||||
| I/O-link | Hub interface (266MB / s) |
Hub interface (266MB / s) |
Hub interface (266MB / s) |
| Features | VIA P4X333 | VIA P4X400 | SiS 645DX | SiS 648 | |
|---|---|---|---|---|---|
| Supported CPU sockets | Socket 478 | Socket 478 | Socket 478 | Socket 478 | |
| Northbridge features | |||||
| Northbridge | VT8754 | VT8754 | 645DX | 648 | |
| Multiprocessor support | No | No | No | No | |
| Front side bus | |||||
| 400 MHz | Yes | Yes | Yes | Yes | |
| 533 MHz | Yes | Yes | Yes | Yes | |
| Memory clock | |||||
| 100 MHz | Yes (400) | Yes | Yes | Yes | |
| 133 MHz | Yes (533) | Yes | Yes | Yes | |
| 166 MHz | Yes | Yes | Yes | Yes | |
| 200 MHz | No | Yes | No | Yes (unofficial) | |
| Asynchronous memory clock | Yes | Yes | Yes | Yes | |
| Memory slots (max) | 3 | 3 | 3 | 3 | |
| Storage support | |||||
| SDRAM | No | No | Yes | No | |
| DDR-SDRAM | Yes | Yes | Yes | Yes | |
| (Dual)-Rambus | No | No | No | No | |
| AGP support | |||||
| 1 x | No | No | No | No | |
| 2 x | Yes | Yes | Yes | Yes | |
| 4 x | Yes | Yes | Yes | Yes | |
| 8 x | No | Yes | No | Yes | |
| Southbridge features | |||||
| Southbridge | VT8235 | VT8235 | SiS 961B | SiS 963 | |
| Disk controller | |||||
| ATA 33 / 66 | Yes | Yes | Yes | Yes | |
| ATA 100 | Yes | Yes | Yes | Yes | |
| ATA 133 | Yes | Yes | Yes | Yes | |
| Number of PCI-slots (max) | 5 | 5 | 6 | 6 | |
| USB ports (max) | 6 | 6 | 6 | 6 | |
| USB 2.0 | Yes | Yes | No | Yes | |
| FireWire | No | No | No | Yes | |
| Other | |||||
| I/O-link | V link (533MB / s) |
V link (533MB / s) |
MuTIOL (533MB / s) |
MuTIOL (1 GB / s) |
As you can clearly see each chip set in their broad differ little. The differences are rather hidden in the details. The newer chipsets from SiS and VIA offer such as the first representative already AGP8x and DDR400, if only unofficially. The most comprehensive Southbridge comes with the SiS648 with its SiS693, which powered the Pentium4 FireWire without additional chip as only southbridge. USB2. 0 has now established an integral part on current motherboards and although the i850E with ICH2 not have USB2.0. 0 has, is a USB2.0 on almost all i850E mainboards. 0 chip being used, which again makes this shortcoming.
The link between North and South Bridge, every manufacturer uses its individual standard. While waiting for Intel to the hub sets interface, VIA with the V-link and SiS its multithreaded used link (MuTIOL) I/o technology. Whether the increase of 533MB / s to 1 GB/s in the SiS648 is really justified and useful, may be doubted but, since not even PCI, USB, IDE, sound and FireWire technology, are expected to use the 533MB/s of the old MuTIOL. A performance gain is in any case not to be expected by increasing to 1 GB / s.
The question is certainly, as it will continue in the future with Rambus support for the Pentium4. SiS has introduced the R658 for Rambus platforms although recently, at least by Intel there but no new Rambus chipsets for the desktop market in the near future. Already the next generation of Intel’s chipsets will include only models with DDR SDRAM support. Already in October, the i845PE, i845GE and i845GV are presented. DDR333 is the magic word for these chipsets then, Rambus is however to see nothing more.
Thanks to a front side bus support 400 and 533 MHz are equipped processors the chipsets listed above all for the new Pentium4. A simple BIOS update that brings the boards at the latest, should be sufficient in all cases.
Compatibility
The Intel disciples are somewhat shaken, which regards the compatibility of their systems to each other. The Sockel423 introduced with the first Pentium 4 held only a year and already was alternated by the Sockel478. This new socket provides somewhat more secure future. At least you can be sure that all processors fit until middle of next year and probably in addition mechanically into the socket, but this can be anything unfortunately already. Already, the fact is that processors on the first Sockel478 mainboards with full time to life can be brought to the new FSB533. A Pentium 4 2.53 GHz would run only with 1.9 GHz (21 x 100 MHz, instead of 21 x 133 MHz). In general all mainboards with old Intel i845, Intel i845D, Intel i850, VIA P4X266A or SiS SiS645 are affected. Partly the chipsets can be while still to operate of the new FSB533 persuade processors, however, this is possible with all mainboards. For the new FSB533, Intel processors presented at that time in May 2002 three new chipsets (i845E, i845G, i850E), were designed for this bar.
So much for the past, how is it now with the new processors from? The good news first: The operation of 2.8 and 2,66GHz Pentium should be ensured to all motherboards based i845E, i845G or i850E. A BIOS update is not even necessary in this case, it is not a good idea however. Plug processor into the socket and off you go.
It is so easy with the 2.5 and 2,6GHz Pentium 4 with his older FSB400 but not. It is possible here a company with the old and new chipsets although it requires a BIOS update here. Our Pentium 4 motherboard, the ASUS P4T533, up the BIOS version 1003 only frequency multipliers up to 24 x offers. Hereby a 2.4 GHz Pentium 4 (24 x 100 MHz) without problems is possible. Since however the multipliers 25 x 26 x lacked and for the processors, so an operation of these models was not possible. Only a BIOS update to version 1004A, which is fortunately for over a week on the FTP from ASUS, managed the hidden multipliers up to 28 x unlock. The two processors can run without problems on older motherboards so if previously a BIOS was recorded, which unlocks the corresponding multipliers.
The table below to illustrate this:
< div >
| Processor | FSB400 mainboard | FSB533 mainboard |
|---|---|---|
| 2, 8 / 133 | No | Yes |
| 2, 66 / 133 | No | Yes |
| 2, 6 / 100 | Yes (with BIOS update) | |
| 2, 5 / 100 | Yes (with BIOS update) |
By the way, the next generation of processors not as readily could already be set to current motherboards into service. The Prescott, who in 2003 will be presented in the second half, a new motherboard needed will be at anyway, because Intel here will increase the frontside bus to 666 MHz (166 MHz).
Overclocking
Even if we still not all too much about the Overclockingfähigkeiten of new 2,8GHz model processors want to types in the scope of this article (another article will follow here soon), is as much betrayed: with air cooling we have stable up to 3,15GHz in Windows and with water cooling, even… more
However, the new nC1 seems basically to bring much potential stepping. We will try to unleash it as far as possible.
Test system
Of course, many new processors demand a new test system. After we the P4T533-C 4 articles in the last Pentium only use for overclocking tests, the big brother of the P4T533-C, the P4T533 with RAID and USB2.0 came in this test now. 0 and 32-bit support Rambus, used. Memory 512 MB of PC4200 Rambus Samsung served us.
The competition from the House of AMD was carried out on the fastest Socket A Mainboard with KT333, ASUS A7V333, tested by us. Here are two 256 MB XMS3000C2 modules from Corsair that had excellent cut in our large Speicherroundup were used as storage. The onboard components of our different testing platforms such as RAID, audio or USB2. 0 have been disabled. In some tests some components are enabled, if for the test necessary. Because our A7V333 with enabled onboard FireWire for no discernible reason was a bit faster without, we have decided here by way of exception, to enable this feature.
As the operating system Windows XP Professional joined as in our usual in the English original version which has been enabled on the respective platforms after the installation. In contrast to previous tests, we have taken this time to the ASUS V8460 with GeForce4 Ti4600, to make sure that the graphics card becomes the braking element in our test course.
To prevent any questions, here is a complete listing of our test system:
- Processor
- Intel Pentium 4 2.80 GHz (FSB533)
- Intel Pentium 4 2.66 GHz (FSB533)
- Intel Pentium 4 2.60 GHz (FSB400)
- Intel Pentium 4 2.53 GHz (FSB533)
- Intel Pentium 4 2.50 GHz (FSB400)
- Intel Pentium 4 2.40 GHz (FSB533)
- Intel Pentium 4 2.40 GHz (FSB400)
- Intel Pentium 4 2.26 GHz (FSB533)
- Intel Pentium 4 2.20 GHz (FSB400)
- Intel Pentium 4 2, 0a GHz (FSB400)
- Intel Pentium 4 1,8A GHz (FSB400)
- AMD Athlon XP 2200 + (2100 + @ 2200 +)
- AMD Athlon XP 2100 +
- AMD Athlon XP 2000 +
- AMD Athlon XP 1900 +
- AMD Athlon XP 1800 +
- AMD Athlon XP 1700 +
- AMD Athlon XP 1600 +
- AMD Athlon 1.4 GHz
- Motherboard
- AMD platform: ASUS A7V333 (KT333)
- Intel platform: ASUS P4T533 (i850E)
- Memory
- 2x256MB DDR333 Corsair XMS3000C2
- 1x512MB RIMM3200 Samsung Rambus
- 1x512MB RIMM4200 Samsung Rambus
- Graphics card
- ASUS V8460 ultra (GeForce4 Ti4600)
- Peripherals
- ASUS CRW 4012A
- IBM IC35LC040
- Driver versions
- nVidia detonator 29.42
- AMD platform: via 4in1 4.42
- Intel platform: Intel inf driver 4.00.1009 + Intel Application Accelerator 2.2
- Software
- Windows XP Professional
Benchmarks
Also in this test, we have spared no effort to highlight the performance of new processors as possible many. Therefore we have expanded on the one hand the benchmarks in the context of this test to three new tests, in addition, we have brought five benchmarks, which previously were used, up to date. To the ‘New’ belongs to a Comanche 4 as DirectX 8 test, as well as Cinema4D XL R7 as the 3D rendering application area. Also tournament is represented also Unreal again in our test.
Updated were id Software Quake 3 arena 1.31, MAGIX MP3 Maker Platinum 3.04, FlaskMPEG with DivX 5.02, spec Vierperf 7.0 and NewTek Lighwave, which took part in the version 7.5 of the test. In total the candidates were able to prove under 16 applications, games, and synthetic benchmarks.
Here a small overview:
- Synthetic benchmarks
- SiSoft Sandra 2002 Pro
- Madonion PCMark2002
- Madonion 3DMark 2000
- Madonion 3DMark 2001 SE
- Games
- Epic Games Unreal Tournament
- NovaLogic Comanche 4
- ID software Quake 3 arena
- Applications
- BAPCo SYSmark 2002
- Winace 2.11
- Lame 3.91
- MAGIX MP3 Maker Platinum 3.04
- FlaskMPEG 0.6 with DiVX 5.02
- SETI@home version
- 3D of rendering performance
- Maxon Cinema 4 d XL R7
- Newtek Lightwave 7.5
- Spec ViewPerf 7.0
Who would like to understand the benchmarks in their home even once, will find much of the above listed test programs with us in the download section.
SiSoft Sandra2002
Before we compete against the team of processors in real-world applications, we want to look once more the theoretical values of the car at this point. To this end we have used Sandra, since both the 3DNow!, extension of processors correctly recognize SSE1 as well as the SSE2 and accordingly also exploited.
Sandra processor test
Sandra 2002 provides equal to two benchmarks, which are to determine only the performance of the processor. This is the Dhrystone benchmark on the one used, which was originally developed by Siemens, to measure the performance of the main processor. On the other hand, the performance of the co processor is determined via the whetstone benchmark. Both tests be carried out without considering the extended such instruction sets.
Sandra 2002 arithmetic
Information on points
|
The first discipline and assay for the new test candidates. In the whetstone as well as in the Dhrystone the new Pentium can say ahead of the competition 4 CPUs clearly. Because here, the pure processing power is measured, the P4 2,8GHz to play processor virtually lossless its clock advantage of 10% compared to the 2,533GHz.
Sandra multimedia test
The Sandra 2002 multimedia test used an algorithm, which is used also when generating realistic natural objects such as mountains and clouds. There is talk here of chaos theory, which was drawn up by Mandelbrot. This benchmark consider the extended command sets of the Pentium 4, or that the Athlon XP. Since the implementation of SSE1 in this partial testing better than that of 3DNow! is, we have the Athlon XP with its SSE1 unit (i.e. 3DNow! professional) work for you. When the Pentium 4, SSE2 was, however, used.
Sandra 2002 multimedia
Information on points
|
The new King from the House Intel can clearly decide for themselves the multimedia benchmark. Generally the use of SSE2 to play here the Pentium 4 unit good.
Sandra memory test
The memory test by Sandra occupies at least 50 percent of available memory. This benchmark to determine the performance of the storage subsystem, as well as the cache. For this, both arithmetic and floating point operations are carried out. As this test very much depends of the platforms, we want to take only a short look at.
Sandra 2002 storage throughput
Information in megabytes per second (MB/s)
|
The classic tripartite Division was certainly to be expected. RIMM4200 is located with its 4, 2 GB/s before RIMM3200 with 3, 2 GB / s and DDR333 with 2 GB / s. In the respective classes, the most powerful CPU each spearheading the ranking. A key advantage of the Pentium 4, namely its possible with Rambus memory throughput, is here beautifully clear.
Madonion PCMark2002
The PCMark 2002 is a comparatively new benchmark, as the name says, designed to measure the overall system performance. For this, he performs a series of tests, which primarily call for the processor and memory. Also the hard disk is checked for their performance there. However, we have given to the publication of the subtotal. Thus, the PCMark provides us with 2002 just a score for the performance of the processor and the memory.
PCMark 2002
|
Additional Computer Articles from "Processors"
- AMD Athlon II X 4 620 and Athlon II X 3 420. Four times please a core for 20 euros each!
- Intel Core 2 Duo E7200. Small big
- Visit to AMD. What's new is in the competition?
- AMD Athlon 64 X 2 5000 + socket AM2. With DDR2-800 for new records?
- AMD Athlon 64 X 2 5000 + socket AM2. With DDR2-800 for new records?
- New benchmarks Pentium 4 FSB 533 MHz. Intel
- Intel Core i7 extreme Edition 980X. Fisted, not waste!
- AMD vs. Intel. View case
- Intel Core 2 Duo E6700 and E6600. Of signs and wonders
- Intel Core i3-2100/2120. Insider's tip for players
Recent Computer Articles
Wood and plasma tv stands: the epitome of style and sustainability in the scenario
Continue Reading | 13 views
Collaboration between Konica Minolta and Komori Corporation for commercial printing market
Continue Reading | 14 views
Income tax is obligatory and pre-registration
Continue Reading | 17 views
Full live interview strategy linking the
Continue Reading | 20 views
Insurance that if divorce becomes a single mother?
Continue Reading | 14 views