Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


Test Drive: Intel Nehalem, Part 2

Welcome back to our presentation of how HP”s new Intel Nehalem-based workstations compare to older workstations when rendering from Adobe Creative Suite 4 (CS4). Briefly, in the last installment, I detailed the tests that I performed, and discussed the results for DV and HDV source materials. This time out, I present the results for DVCPRO HD, AVCHD, and Red and share how the Z400 and Z800 performed with Hyper-threaded Technology (HTT) enabled and disabled.

Ready? Let”s jump in.

Table 1. DVCPRO HD results.


DVCPRO HD has four times the data rate of either DV or HDV, though it”s an intraframe-only format so it”s easier to process than HDV or AVCHD. In fact, the easiest way to think of DVCPRO HD is as four streams of DV, one for each quadrant.

Related Links

Test Drive: Intel Nehalem, Part 1

A few months ago, I ran some Adobe Creative Suite 4 (CS4) benchmarks on different computers that isolated how CS4 performed with formats ranging from DV to Red. Now that Intel”s Nehalem processor is upon us, those numbers are obsolete…

Configuring Your System for CS4, Part 1

This month, I”ll review how the number of processors and the amount of RAM impact encoding performance, and then find other, non-CS4-related fish to fry…

CS4 and 64-bit Systems, Part 1

So there I was, testing Adobe Creative Suite 4 (CS4)”s AVCHD compatibility. I created a simple project, about 4 minutes long, two picture-in-picture overlays with simple rotation and color correction…

Recall that the shorter test was relatively effects-heavy, with an Adobe chroma key effect added via Dynamic Link. On the short project, where throughput was less critical than processing power, the Z800 and xw8600 posted nearly identical scores, with the single-core Z400 significantly behind. Since DVCPRO HD is so bulky, it”s not surprising that the eight-core, 32-bit xw6600 bogged down, much more so than the xw4600 four-core 32-bit workstation. If you”re producing on a 32-bit workstation, don”t forget that Dynamic Link is an option, not a requirement. You may get overall faster results by rendering out your After Effects project first, and importing the result into Premiere Pro, or Encore for that matter.

On the longer project, the Z800 was the perfect combination of throughput and processing speed, rendering in half the time of the xw8600. With only 6GB of RAM, the Z400 really suffered, bested by the xw8600 with 16GB of memory. The bottom line is that if you”re editing DVCPRO HD, 8GB should be the minimum RAM configuration, with 10GB or 12GB a worthwhile investment.

Again, if you”re running a 32-bit workstation with room to expand your RAM, you”ll find upgrading to 64-bit Windows an inexpensive but highly effective upgrade. It’s tough to imagine that many DVCPRO HD producers are working on a single- or dual-core system, but if you are, you should have little trouble justifying the upgrade.

Table 2. AVCHD results.


AVCHD was the original memory hog that got me started comparing 32-bit to 64-bit systems a few months ago, so I wasn”t shocked that all 32-bit systems were outclassed in the memory-intensive first project, which involved the After Effects chroma key effect inserted via Dynamic Link.

On the short test, the Z800 outperformed the xw8600 by only a small margin, though the advantage grew significantly with the longer project. While the results weren”t quite as impressive as with DVCPRO HD, a dual-Nehalem-type workstation such as the Z800 looks to be a good investment for AVCDHD editing, both compared to older-technology eight-core systems and single-Nehalem-based workstations such as the Z400.

Moving further to the right, it”s interesting that the four-core xw4600 workstation outperformed the eight-core xw6600 in both tests, proving that more cores aren”t better when working with a memory intensive format like AVCHD on a limited RAM system. If you have the room to expand your RAM to 8GB, upgrading to 64-bit Windows is a total no-brainer for AVCHD editors.

Table 3. Red results.


In the memory-intensive first project, the two computers with the most RAM easily outperformed the others—including the Z400, which was seriously hobbled with only 6GB of RAM. Irrespective of processor, if I”m editing Red footage, I”m thinking 16GB is the minimum, or I”m eschewing Dynamic Link and other techniques that push memory usage.

On the longer project, there really doesn”t appear to be a wonderful solution, as our fastest time was just less than 5 hours, or about 30X realtime, for our 10-minute project. Though you would think that the Z800”s throughput would trump the xw8600 in this test, when rendering is so slow, throughput isn”t a bottleneck. Bottom line is that processing Red is very inefficient from the processor”s perspective, so even the world”s fastest desktop CPU doesn”t stand out.

On a positive note, Red also doesn”t appear to be that memory-intensive without Dynamic Link in action, so if you”re editing simple, long-form projects on an older computer, your performance is actually pretty reasonable. After all, who would have guessed that the Pentium D-based xw4300 would be the most competitive on a percentage basis when editing the highest-definition content? That said, in this case, though the xw4300 was only 172 percent behind, this translated to more than 7 hours of additional rendering time, which is plenty meaningful.

I don”t know if there”s anything Red or Adobe can do to make working with Red footage more efficient in CS4. Until they do, however, on longer-form projects, you”re not going to see much benefit from a Nehalem-based system.

Table 4: Rendering times, with and without HTT.

More isn”t Better (at least not always)

As mentioned on page 1, I tested the Z800 and Z400 with HTT enabled and disabled, and I included the fastest times in the comparisons above. Table 4 presents these results. To explain, a positive percentage in the HTT column means that rendering with HTT enabled was slower, while a negative time means that enabling HTT reduced rendering time. Note that some of the swings were meaningful, led by the 32 percent slowdown that enabling HTT induced when rendering Red footage with the Z400.

It”s tough to draw any logical conclusions from the results since they”re often contradictory. For example, enabling HTT for the short AVCHD project slowed the Z800”s rendering time by 20 percent, but speeded rendering in the longer project by 15 percent. The bottom line is that you shouldn”t assume that HTT will accelerate your rendering times. If your project types are relatively consistent—say, the same 30-minute show each week—you should test with HTT enabled and disabled to determine which configuration is faster.


In case you haven”t noticed, I really like the performance of HP”s new Nehalem-based workstations, which would appear to pay for themselves in saved rendering time for most HD formats in a just a few months. However, if you don”t have the spare coin right now and you’re looking for a less-expensive solution, try adding RAM to your system—especially if you”re editing AVCHD, DVCPRO HD, and Red footage. Of course, you”ll need to be running 64-bit Windows to do so.

I know I”ve said this before, but I can”t make the point too many times. That is, if you”re running on 32-bit Windows with CS4, please stop what you”re doing right now and check how much RAM you can add to your system. If you can install up to 8GB or so, upgrade to 64-bit Windows, Vista or XP, as soon as possible. I guarantee that your rendering times will improve, your blood pressure will go down, and you”ll have a more positive attitude about life. You and everyone around you will want to thank me personally for encouraging this decision.

If you can”t expand your RAM configuration significantly and you have to work in 32-bit Windows, you should not use Dynamic Link or run other CS4 applications while rendering in Adobe Media Encoder. This pushes your Windows configuration to the limits, slows rendering dramatically, and can cause system instability.