by FastSite
12/03/2002 | 12:00 AM
Today we will try to get more intimate with the new IDE drive from IBM. It belongs to the 180GXP product line and is code-named Vancouver2. So what's so good about this particular hard disk drive?<%BANNER[article]%>
Firstly, the drive is built with 60GB platters. With the three-platter design, it yields the maximum capacity of 180GB.
Secondly, the platters, according to the press release, are made of aluminum, not glass. That is close to sensation, to tell the truth, as IBM had been using glass platters for years and had been very proud of it. And now the company changed its course to just the contrary…But, as the wise man says: "to betray timely means not to betray, but foresee". :)
Thirdly, this is the first HDD model from IBM to use hydro bearings! A second sensation in a row!
One more trump in IBM's deck is their house technology called "tag 'n' seek". Its concept is quite familiar, though: the HDD controller arranges the commands it receives in the most optimal order (as the controller thinks).
As we are going to discuss the technology in a separate article, we won't dwell upon it here and suggest coming over to the drives we tested this time.
There is an all-stars show we have managed to gather for this comparison with the IBM 180GXP hard drive. Of course, we used the results of our tests for Maxtor DiamondMax Plus 9 and Seagate Barracuda ATA V. These are the immediate rivals of the new IBM's drive as they feature the same 60GB platters. Maxtor will also star in both UDMA100 and UDMA133 modes.
And to realize how big the progress in hard drive making has become lately, we will take IBM 120GXP (80GB) and include it in our test session as well.
The table below lists the main features of IBM hard disk drives of the last two generations:
There are two features in the new HDD specs that need commenting on. Firstly, as the capacity of the high-end drive in the 180GXP line stepped over 137GB, IBM had to re-design the electronics of the new drive to add ATA/ATAPI-6 support. This in not UDMA133, although some tricky firms started touting "IBM 180GXP U133".
Secondly, IBM 180GXP features hydro-bearings and thus low-end models are claimed to produce really little noise.
But let's meet the new IBM 180GXP HDD:
Seems familiar, doesn't it? Yeah, that's exactly the design of the Vancouver LP model, which appears the last thing I expected to discover here, actually. Well, maybe it will be used in one-platter Vancouver2 drives only. And maybe the difference in access time between the top and the low-end models is somehow connected with the case design?
The snapshot of the PCB shows one more difference the new IBM drive can boast compared with its competitors: the dedicated read/write channel chip. In the HDD specs, IBM claims the external read/write channel is the most reliable solution. Well, let's see.
We used the following testbed:
We involved the following software:
Before the tests the AAM register of all HDDs was set to OFF position (FAST mode) via IBM Feature Tool Utility.
The IBM IC35L060AVV207-0 HDD we tested had the V220A60FB0 firmware. Among the strange things we would mention that SMART support was disabled by default.
For WinBench tests all the drives were formatted in FAT32 and NTFS as one logical drive with the default cluster and for copy tests the drive was formatted as two logical drives of the same size. The tests were run four times each, the average result was taken for the diagrams. The drives didn't cool down between the tests. The tests in Intel IOMeter were run in SequentialRead, SequentialWrite, DataBase, WorkStation and FileServer patterns. If you are looking for the detailed description of these patterns, please see our previous articles.
Following the good old tradition, we will start with HDTach.
The first thing to consider is the high access time, which is quite contrary to what we saw in the last IDE drives from IBM. It's 13.6ms (and should be ~13 (8.8+4.17)…). We can't remember an IBM drive diverging from its specs so much. :(
The maximum read speed however, nearly notched 60MB/sec, which is a rather high result for a drive with 7,200rpm spindle rotation speed.
Let's check the average read and write speeds of the new drive from IBM:
As we see, they are much higher by IBM 180GXP than by competitor products.
Winbench99 is the second test we will take a look at. The previous HDD model from IBM, 120GXP, proved fast here and even rivaled a drive from WD with a large 8MB cache buffer. What has the new one got to say?
The linear read speed of the new drive is about the same as that of the predecessor: 55MB/sec. We say "about" as the IBM drive drew a "jugged" graph and the exact read speed at the densest zone may be a little higher in the end.
As for the results in integrated tests, we'd better watch them on a diagram:
And the results are curious, really. IBM 180GXP lost to the previous generation IBM drive and Seagate Barracuda ATA V in Business Disk WinMark! That's a rare thing to see.
In High-End Disk WinMark the new IBM drive regained its leadership, though. But look, it's just 100KB/sec better than the 120GXP. Well, maybe it's the limit for a 2MB cache-buffer HDDs?
Let's check it out in NTFS.
There are two definite leaders in NTFS. In the Business Disk WinMark, Seagate showed its best, while IBM 180GXP is ahead of all in the High-End test. Curious, but the new IBM drive again loses to 120GXP in the Business test in NTFS.
There are two more diagrams based on the results obtained in WinBench. The first one shows average access time to the random sector compared to the average access time as measured in HDTach.
Well, we see that IBM 180GXP takes the last place among all the competitors in terms of access time.
But in linear read speed, IBM 180GXP looks quite well. Moreover, the read speed of 180GXP at the beginning and end of the disk is much faster than in competitor products.
The WinBench section of the review winds up with the linear read graph of IBM 180GXP (Click Graph to open).
So, we know now that the new hard drive from IBM features high data density, but also high access time. It's not good, as the following tests require the HDD to move its actuator fast. Let's not forget, though, that the good firmware algorithms affect the overall result in these tests, too.
The DataBase pattern will help us to check the drive's ability to sort up read and write requests and the quality of the read ahead and lazy write algorithms.
When the request queue depth is equal to 1, IBM 180GXP isn't impressive at all. Although it outperformed Maxtor DM Plus 9 in random read and lost only to IBM 120GXP in random write, it lost to the drives from Seagate and Maxtor in most mixed modes.
It's clearly seen that the same "hand" designed both IBM drives: their graphs are very similar. So we may suppose that the difference in their results is due to deferent average access time values only.
As the request queue depth increases, the situation changes drastically. Thanks to great job of the HDD controller on sorting requests and optimizing their processing order, IBM 180GXP surpasses Seagate and Maxtor HDDs in most modes. Note also, that as write operations share increases, the new IBM drive is steadily closing to 120GXP.
The IBM drives easily leave all others behind at this request queue depth. Seagate Barracuda ATA V could only compete with IBM 180GXP in the RandomRead mode.
The results of the DataBase pattern allow us to say that the new hard drive from IBM keeps the brand "gene" of IBM, but, regrettably, turns to be slower than IBM 120GXP.
Here we are going to check the new drive's ability to process sequential read and write requests:
The table is all right, but the diagram is prettier, of course:
An unpleasant fact for IBM: the new drive lost the secret of high-speed processing of small data blocks, which IBM 120GXP used to boast.
The same picture at writing. Although the 180GXP graph, displaying dependency of data processing speed on the data block size, is most beautiful, we can't omit the fact that 180GXP processes small data blocks the slowest of all.
After the heart-breaking results we have got in synthetic patterns, let's go over to the patterns that try to emulate real-life applications. The first of them is WorkStation pattern.
We see the new IBM drive losing to all other drives under smaller workloads (small request queue depths). And as we calculate the overall rating in this pattern according to the formula:
Performance = Total I/O (queue=1)/1 + Total I/O (queue=2)/2 + Total I/O (queue=4)/4 + Total I/O (queue=8)/8 + Total I/O (queue=16)/16 + Total I/O (queue=32)/32,
where the HDD speeds under smaller workloads have bigger weight, it's no wonder IBM 180GXP got the worst total rating here:
And the highest rating was won by IBM 120GXP, the previous generation drive from IBM. The new drive from Seagate, Barracuda ATA V, took the second place, but still slightly behind the first.
These patterns, as their names suggest, emulate disk subsystem workloads typical of File- and WebServers.
As the server may encounter any workload with equal probability, the "performance rating" in these patterns is calculated by determining the average of Total I/O values:
IBM 120GXP is still on top, while IBM 180GXP literally squeezed into the second place. Look how close the Barracuda ATA V is!
The last test in our session is going to check HDD performance during creating, reading and copying files.
Showing our respect to the NTFS, which is getting more and more popular (due to MS' prayers of course), we will start with this file system:
Maxtor is fastest writing file sets onto a blank HDD, while the two IBM drives fight for the second position. The newer one, 180GXP, is better with large files, and the older one, 120GXP, - with smaller files (Windows and Programs patterns).
When reading the file set from the drive, new IBM drive breaks loose from the others as it comes to large files. But when it comes to smaller files, Maxtor's drive takes the lead.
The results of copying files inside one partition are rather messed up and can't be described in two words... :)
But the leaders are the same: IBM and Maxtor drives. Note also, that Maxtor HDD shows exceptionally high speed at copying a really big file (~500MB).
When copying file sets from one partition to another, there appeared two leaders: Maxtor did best on large files and IBM 120GXP - on small files.
A courtesy to NTFS has been made, let's get over to FAT32! :)
A nice thing to have something permanent in our unstable world: Maxtor drives are again the best at writing a file set. And the size of the files doesn't actually matter.
IBM 180GXP reads files faster than the rivals.
Well, IBM 180GXP was also the fastest at copying files inside one partition, except CD image files where Maxtor managed to beat it.
Maxtor and IBM drives are again best at "far-copy" and the first one again favors large files, while the latter - smaller ones.
FC-Test showed that IBM 180GXP is fastest of all in reading a file set from the drive. No wonder as it boasts the highest linear read speed of all HDDs present.
The temperature was measured after the HDDs had been actively working in the system case for an hour and a half.
As the last generation Maxtor drives display too high temperature values in the SMART area, we carried out a little investigation to find the thermal diode in them. After an hour and a half of continuous work, the HDD was taken out of the system case and explored with an infrared thermometer to find "hot" spots. We found three of them (it was known beforehand, although theoretically). The warmest spot of the HDD is the electronics PCB, namely the motor-controlling chip. The second "hot" spot is where the actuator is fastened to the PCB. The third "hot" spot is the shaft & bearing.
The table below lists the temperature values taken by means of Dtemp utility and an infrared thermometer at the upper cover of the HDD case and electronics PCB (max temperature).
As we see, the thermal diode of Maxtor HDD is situated close to the motor-controlling chip.
The second discovery we made was the low heat dissipation by IBM 180GXP! We remember IBM drives used to heat up a lot, and here we suddenly see such a surprise! Well, the transition to hydro-bearing positively affected the HDD temperatures. Moreover, the new case design adds to better heat take-off, for sure.
The Seagate drive was somewhat at uneven terms with the competitors as it featured two-platter design and thus heated up more. The same was valid for IBM 120GXP.
The PCB temperature by Seagate Barracuda ATA V, as you understand, couldn't be measyred due to the protective SeaShield cover.
Noise level of the HDDs was measured subjectively, that is, "by means of a human ear" :).
Barracuda ATA V is the quietest here as it is one of the most noiseless drives we have ever met. Maxtor drive is the second: it is nearly as quiet as Seagate Barracuda ATA V. IBM 180GXP takes the honorable third place. By the way, the difference in the "noise" produced by the HDDs had to be measured by pressing the ear to the upper cover of the working HDD. Where is this world going to?
It's the first time that we see the tradition of IBM to be no longer valid. Remember that their drives with N data density always came up much later than those of the competitors, but still were the fastest. IBM 180GXP hit the market the same time as HDDs from Maxtor and Seagate, but its performance in some tests turned to be no better than that of the previous generation IBM drive, 120GXP.
Considerable increase in the track density per inch (from 56700 by IBM 120GXP to 72000 by IBM 180GXP) resulted in higher average access time of the new drive and negatively affected its performance in some tests.
Poor results of the new drive in tests with sequential requests must have been caused by the "raw" firmware and will hopefully get improved soon.
According to the specs, IBM drives with larger storage capacity will have lower access time (another case?) and their performance will be higher than that of the drive reviewed today. And of course we are all looking forward to IBM's HDDs with the 8MB cache-buffer. :)
