Almost a year ago we managed to take a closer look at a HDD with 20GB platters. It was IBM DTLA 305020. This hard disk drive had a revolutionary influence on our mind because it was for the first time that we saw a HDD with 5400rpm rotation speed outperform the one with 7200rpm. This incredible (at that time) data density made up for a relatively low spindle rotation speed when working in common, so called "business applications". Besides, the read/write heads of this hard disk drive were also moving quite fast. IBM storage solutions have always been a bit more expensive than the products from their competitors. But at that time this 1GB drive was a really attractive offer.

But time passed on and 20GB platters were no longer a unique feature of IBM HDDs. The company's competitors also learnt to use them. The first competing hard disk drive with 20GB platters we got hold of was Seagate U5, however, as we have already seen, it failed to surpass the IBM guy. But, this leadership didn't last long. As soon as there appeared new WD Protege HDD, IBM DTLA had to share. The young competitor very quickly beat the veteran in business applications.

Then the hard disk drives with 20GB platters started storming into the storage market like from a horn of plenty. However, they featured 7200rpm spindle rotation speed. These were the babies from Seagate, WD, Quantum. They quickly raised the performance bar quite significantly (see here for details) and increased the gap between the drives with 7200rpm and the best drives with 5400rpm. At that time we waited impatiently for a solution from IBM rotating at 7200rpm, but in vain: it didn't come out (btw, the one from Fujitsu also didn't come then). Therefore, IBM 75GXP family (also known as IBM DTLA) had to stand the competitors' attacks. And finally the rivals' performance got so close to that of IBM 75GXP, that it was even defeated in some cases! So IBM's leadership was no longer that indisputable…

Now we are holding the newest IBM baby - IBM Deskstar 60GXP (IC35L040)...

Closer Look

The drive hardly differs that much in the outlook, the cover seems to be absolutely identical:

IBM Deskstar 60GXP (IC35L)		IBM Deskstar 75GXP (DTLA)

The electronic part seems to have undergone some modifications though:

IBM Deskstar 60GXP (IC35L)		IBM Deskstar 75GXP (DTLA)

Note that the PCB of the IBM IC35L drive doesn't have a cache-memory chip any more! It is either integrated into the main DSP or moved to the reverse side of the PCB (which is actually very unlikely to be true). We decided not to disassemble the HDD, since it was provided for testing and not for autopsy :-)

The electronic board got somewhat smaller, as you may see, and the DSP chip lost its protective lid. The case construction has also been changed a bit: do you see those strengthening ribs around the center?

But the most interesting thing is the small details, of course. To tell the truth, we were very upset when we saw the following sticker on the new IBM Deskstar 60GXP HDD:

IBM Deskstar 60GXP (IC35L040)		IBM Deskstar 75GXP (DTLA 307020)

Well... This had to happen some day. We only hope that the quality of this product will be as good as that of the predecessors...

Brief Technical Specs

As for the other testing participants, you may find their specs in the corresponding reviews on our site.

Judging by the listed parameters, the new hard disk drive will be not in the least worse than IBM Deskstar 75GXP (DTLA 307020). However, we are still very much puzzled how on earth IBM managed to retain the full stroke seek time unchanged while the number of cylinders grew…

There are a couple of other things worth mentioning separately. The first one is the data density. The read speed from the disk surface in the first zone makes 494Mbit/sec (DTLA featured 444Mbit/sec), which is just the same as by Seagate Barracuda ATA III. It means that theoretically, there should be two leaders from the read speed point of view.

The second thing, is the number of recording zones. Now it is equal to 18 (compared with 15 by DTLA). In other words, IBM tries to squeeze the maximum out of each platter zone, reducing the number of tracks to the minimum.

And the third thing, we would like to point out is the noise level, which remained the same as usual:

• 1 disk - 2.9 Bels
• 2 disks - 3 Bels
• 3 disks - 3.1 Bels

Of course, these numbers correspond to the fast mode, since this hard disk drive is equipped with AAM system, just like IBM DTLA.

Testing Environment and Methods

• Intel Coppermine 600MHz CPU;
• ASUS CUBX-E mainboard, 1007A BIOS;
• 2 x 128MB PC133 SDRAM by Hyundai;
• Matrox Millennium 4MB graphics card;
• Windows 98/Windows 2000 Pro.

In order to check the HDDs performance in different UDMA modes, we had to use the following controllers:

• UDMA33: the controller integrated into i440BX chipset;
• UDMA66: Promise Ultra66 controller;
• UDMA100: integrated Promise Ultra100 controller.

For the Promise controllers we used the drivers ver. 1.60 (build 33).

The disk drives were connected as Master-units to a separate IDE-channel. DMA support in Windows was enabled. We used FAT32 and NTFS file systems to format each of them as one logical drive of the maximum size with the default cluster. All the tests were run 4 times and then the average results were taken for the diagrams. The HDDs didn't rest for cooling down between the tests.

Here are the benchmarks used:

• Windows 98:
  • WinBench 99 1.2
  • Adaptec Threadmark 2.0
• Windows 2000
  • WinBench 99 1.2
  • HDTach 2.61
  • IOMeter 1999.10.20

In this review we will compare the performance of all the today's hard disk drives with 7200rpm spindle rotation speed and 20GB platters. They are: Seagate Barracuda ATA III, Western Digital Caviar WD200BB, Quantum Fireball Plus AS and Maxtor DiamondMax Plus 60. Also we included the previous IBM solution - IBM Deskstar 75GXP (DTLA 307020) and a representative from a new IBM Deskstar 60GXP family - a 40GB HDD. Unfortunately, we didn't manage to get a 20GB model for our investigation, that's why we had to break our principles, but IBM 60GXP is really worth it :-)

And seriously speaking, we will definitely do our best to take a close look at the 20GB model as well, as soon as it is available.

Performance

Access Time

Well, we are very much excited to see if the claimed average seek time of IBM Deskstar 60GXP (IC35L) equal to 8.5ms corresponds to the reality:

Let's find it out: the smallest average access time we managed to get equals 12.5ms. Now we subtract half the platter rotation time, i.e. 4.17ms, and get the average seek equal to 8.33ms. Just what they promised!

However, when we measured average access time of the "silent" IBM IC35L040, the benchmarks showed different results: the results obtained in WinBench99 appeared 1.5 times higher than those in HDTach. Which one was the most credible then? Judging by the previous tests, HDTach was closer to the true performance, but we still had some doubts about it. To solve this matter, we had to resort to Drive 1.0 program. This arbitrator showed 22.2. So, we concluded that Disk Inspection Test from WinBench99 package didn't measure the performance correctly for the drives working in the "silent" mode.

The regular (i.e. "non-silent") IBM IC35L040 HDD proved the fastest among drives with 20GB platters, which was actually quite predictable.

Data Density

Well, to our great disappointment no sensation happened. The fastest reading from the platter belonged to Seagate Barracuda ATA III, as usual. But IBM IC35L040 won a very good second position having yielded only 700KB/sec to Seagate, which is a really good result. However, the linear read graph of IBM IC35L040 in UDMA33 mode made us quite concerned. In UDMA66/100 protocols the graph seems to be a little bit "trembling" while in UDMA33 something really terrible happens to it. Maybe it is caused by the significant difference between the high platter read speed and lower buffer read speed. But if there is more water getting into the pool than they take from it, then they should in no case suddenly run out of water. Let's imagine a situation when the pool is filled up to the very edges. In this case a man or an automatic system watching the process should close the taps. The water will continue getting out of the pool via a special tube and then (attention!) it is essential that the filling taps are open again at the right moment, not to let the pool get completely empty. If it is a continuous stream and the coming stream is larger than the leaving one, then there shouldn't be any problems. But the HDD features a non-linear data structure and can't always provide the next data portion rapidly enough. This may be one of the reasons for the linear read graph to be like that. But, enough for the pools and taps, we continue with hard disk drives...

As we see, the data density on the outer tracks of IBM IC35L040 drops almost twice compared to the density on the inner tracks. We haven't yet seen anything like that before. However, there are no causes for concern, since the advice to locate the workspace closer to the first tracks is still valid.

HDTach 2.61

Again we are mostly interested in the average read and write rates:

As you can see, IBM IC35L040 HDD appeared a bit slower than Seagate Barracuda ATA III but left behind Maxtor DiamondMax Plus 60 in UDMA66 and UDMA100 modes. In UDMA33 its performance turned out quite mediocre, which signifies that it isn't well-optimized for this protocol. However, Seagate Barracuda ATA III also doesn't like this interface that much. It looks as if IBM and Seagate's engineers decided not to waste effort trying to optimize the algorithms for UDMA33 and preferred to focus on more up-to-date ones. Well, they might be right.

Hm, the leader has suddenly changed! IBM IC35L040 won the laurels here. Then follows Maxtor baby. The last one is the old buddy IBM DTLA.

In synthetic benchmarks the new IBM solution looks very attractive, leaving behind its DTLA predecessor quite significantly. As you remember, in WinBench and IOMeter IBM DTLA used to simply destroy its competitors. Now let's see what about IBM IC35L040.

WinBench99 for Windows98

Well, here are the results obtained in the benchmarks run under Win98.

All in all, IBM Deskstar 60GXP (IC35L040), Maxtor DiamondMax Plus 60 and Quantum Fireball Plus AS proved very close to one another. Although the picture was a bit different for every UDMA protocol, we would still regard IBM as a winner here.

Well, there is no need in being subjective here: the leadership indisputably belongs to IBM IC35L040. And the second prize was also taken by an IBM solution - IBM DTLA.

Adaptec Threadmark 2.0

Well, what does the situation look like here?

Maxtor product is running ahead of all here, and IBM IC35L040 fights with Seagate Barracuda ATA III and IBM DTLA for the second position (and is very likely to win).

WinBench99 for Win2000 (FAT32)

When we ran the tests under Windows98 we failed to single out a definite leader in business applications. That's why we had to take all the responsibility and to give the victory to IBM IC35L040. It is quite possible that the results appeared pretty unstable because of the 24bit OS core. Well, let's see what we will get in Windows2000.

In UDMA66 and UDMA100 modes the results are beautifully stable, which makes us believe that the surprising behavior of the HDDs in Win98 was caused either by the instability of the operation system or by the driver problems.

The results obtained here allow us to state that IBM IC35L040 is almost 14% ahead Maxtor drive, which boasts very similar features. The gap between the leader and its predecessor, IBM DTLA isn't big at all (about 4%), but it is undeniable. In silent mode the performance decreases a bit, right down to that of IBM DTLA.

This is a real SHOCK! Like a bolt from the blue! The performance increase is just unbelievable. Well, we do understand that the data density is higher, but it is only 11% higher in the densest zone, and the performance grew by the good 12%! :-)

If you remember, in the very beginning of our review, we told you that we didn't see the cache chip on the drive and supposed that it was integrated into the chip (take, for instance, graphics chips, which start little by little acquiring more and more caches). When we analyzed the benchmark results obtained, we doubted this supposition. That is why we pulled ourselves together and removed the electronic board.

Here it is! The cache microchip is right there. That is why IBM IC35L040 most probably owes its cool performance increase to higher data density and the enhanced DSP and its firmware.

WinBench99 for Win2000 (NTFS)

Here come the graphs:

In UDMA66/100 IBM IC35L040 HDD is still ahead of its rivals, but in UDMA33 it yielded to Quantum Fireball Plus AS (which proved very "robust" to protocol changing) as well as to IBM DTLA.

If we calculate the advantage of IBM IC35L040 over its closest competitor with 200GB platters in percents, we will get an impressive number: 19%!

We are no longer surprised to see IBM IC35L040 being an indisputable leader. In UDMA100 this drive leaves WD200BB almost 17% behind.

So, in WinBench99 we got a new absolute winner among IDE hard disk drives. It is for the first time that the result obtained in High-End Disk WinMark exceeds 30,000. As far as we remember, only Quantum Atlas 10K II with 10,000rpm rotation speed and Seagate Cheetah X15 proved capable of beating this record (see our Seagate Cheetah 36XL HDD Review).

Now let's turn to IOMeter to see what IBM Deskstar 60GXP (IC35L040) is capable of.

Intel IOMeter

We would like to start with comparing the fast and the silent models of the new HDD, and then we will pass over to a comparison with the competing drives.

We have already devoted enough time to the details and principles of Intel IOMeter benchmark. So, you may consult our previous storage reviews as well as this article on StorageReview. And we continue.

The diagram below shows the dependence of Total I/O on the load for Fileserver access pattern. We compared IBM IC35L040 with IBM DTLA 307020 in both: silent and normal modes:

No doubt, IBM IC35L040 inherited all the best from its predecessor. The graphs obtained in fast modes are almost parallel. New IBM IC35L040 appeared a bit better than its forefather and its graph is situated a bit higher on the diagram. It is interesting that in case of high workload the graphs for both modes intend to meet in one point again. Moreover, silent IBM IC35L040 turns out more aggressive than silent IBM DTLA. It is most likely to be connected with higher clock frequency of IBM IC35L040's chip.

And now let's compare the Sum Total value of both HDDs. The Sum Total is calculated taking into consideration the results obtained with Light, Moderate and Heavy loads:

Well, the new IBM Deskstar 60GXP (IC35L040) proves the best again and shows perfect results in this server benchmark.

Heat and Noise

We measured the heating with the external thermal diode connected to the mainboard. IOMeter takes over 2.5 hours to run just once and the HDD works quite tensely all this time that's why we may regard it as normal thermal conditions for the drive. We put the HDDs into a PC case, without any extra coolers installed. The room temperature was around 22 degrees Centigrade. As soon as the test came to an end we restarted the system and took the temperature from the system BIOS (1007A). Maxtor and IBM drives were tested with AAM (Automatic Acoustic Management) modes set to normal, i.e. fast mode.

We tested both: single-platter and dual-platter HDDs and no wonder that dual-platter models heated greater. However, the surprising thing was to see a dual-platter IBM IC35L040 dissipate less heat than a dual-platter IBM DTLA. The difference made 5 degrees! That is why we really hope that the new HDDs won't suffer from overheating.

As for the noise level, we have to confess that it is pretty hard to make any evaluations without proper equipment. That is why while our "quiet room" isn't ready yet, we have to rely solely on our impressions.

When we tested hard disk drives with WinBench99 benchmarks we didn't hear any noise. We mean that the spindle was hardly hearable against the background of a working cooler. When running the IOMeter tests, the HDD is overloaded so heavily that the entire PC case starts making some troubling noises. The heads also start making a really noticeable noise. Though we described the noises made by the DTLA's heads as some kind of squeak, in case of our today's hero it sounds more like rustling.

Conclusion

New IBM Deskstar 60GXP hard disk drive showed brilliant performance in all the benchmarks, file and synthetic ones. Sometimes it left its competitors with similar features up to 20% behind, which is a really impressive achievement for a HDD of the same generation as its counterparts.

However, we would like to say that high performance of IBM Deskstar 60GXP (IC35L040) in file tests may result from the fact that we compare HDDs of different storage capacity (we have already pointed out some time ago that WinBench results are dependent on the cluster size, while the latter depends on the capacity of the HDD). Unfortunately, we don't have any IBM IC35L HDDs of other capacity at our disposal right now, however we are really willing to get hold of them…

So, let's consider the results discussed in this review preliminary and on our part we promise to carry out more tests for a 20GB model as soon as we get it.

Among he most unpleasant issues we would like to mention the cache-buffer microchip moved to the inner side of the electronic board. In this case the microchip can get overheated and hence the data stored in the cache can be lost. Maybe our suspicions are totally ungrounded, but… To tell the truth, we wouldn't recommend fastening this HDD inside the PC case with all the electronics turned upwards and without any extra cooling on that side.

In general, some time ago we thought that the hard disk drives with this set of features were nearly ideal now, and that it would be hardly possible to get to a new level without some drastic changes, such as, for example, increase in data density, cache-buffer size, rotation speed, etc. Now that we got acquainted with IBM Deskstar 60GXP we changed our mind and are no longer making conclusions like that.

And what are we up to in the nearest future? First of all, we do hope that one day the last HDD with 20GB platters, which we haven't yet tested, Fujitsu MPGxxxxAH-E, finally reaches Russia and we will be lucky to get it. Secondly, there should very soon appear the notorious WD800BB from Western Digital. This HDD features 3 platters, 27GB each and has every chance to beat IBM Deskstar 60GXP (IC35L040). And finally in summer we expect some new IDE solution from Seagate with 10,000rpm rotation speed. We wonder what they will call it: Cheetah IDE or Barracuda 10K? :-) All in all, summer promises to be a hot period…