Sign in with
Sign up | Sign in
Intel Xeon E3-1280 v2 Review: Ivy Bridge Goes Professional
By ,
1. Ivy Bridge Finds Its Way Into Servers And Workstations

Not long ago, we took a look at Intel’s Xeon E5 family in Intel Xeon E5-2600: Doing Damage With Two Eight-Core CPUs. In that story, we saw the company expose its Sandy Bridge-E design the way its architects originally intended: armed with eight cores, 20 MB of shared L3 cache, and QPI links cranking away at 8 GT/s. That was a far cry from the desktop Core i7-3960X we reviewed previously, neutered back to six cores and 15 MB of L3 cache—particularly since our Xeon E5-based platform was running in a dual-socket configuration in Altered Beast mode.

Now, Intel is replacing its entry-level Xeon E3s, formerly based on the Sandy Bridge architecture, with models that employ its Ivy Bridge design.

There aren’t any dormant processing cores or blocks of shared L3 cache to make the new Xeons really pop this time around. Intel instead relies on a couple of unique business-oriented features and creative pricing to make its Xeon E3s shine.

How’s this for an example? Right now, Newegg is selling the Core i7-3770 (with a locked multiplier) for $320. A Xeon E3-1240 v2, which runs at the same 3.4 GHz, doesn’t include processor graphics, and bears a 69 W thermal ceiling, goes for $280. Same speed, no useless graphics, lower power consumption, and $40 less? Color us intrigued.

The Benefits Of Workstation-Class Hardware

Today, however, we’re looking specifically at Intel’s Xeon E3-1280 v2, which is higher up in the stack. Just 100 MHz faster than Intel’s quickest desktop-oriented model, Core i7-3770K, E3-1280 v2 sells for a pricey $612. The -3770K is a much more palatable $332. How do you justify the Xeon over the Core i7 when you know they’re both based on the same Ivy Bridge architecture?

That’s a difficult argument for an enthusiast to make. But in some environments, the difference in pricing is truly inconsequential compared to the few benefits enabled by the Xeon. Compared to the desktop parts, Intel’s Xeon E3s support ECC-capable memory, for example, catching and correcting memory errors that could either take a business machine down or affect critical data.

These new Xeons also enable four additional lanes of third-gen PCI Express connectivity. A chip like Core i7-3770K exposes 16 lanes, total. The Xeon E3s offer 20, which can be used as single x16 and x4 links, or across two x8 slots and one x4 interface. The former configuration would work well in a workstation with discrete graphics and an add-in storage or networking controller. The latter is decidedly better for server setups, supporting more high-speed devices over an 8 GT/s bus.

Some of the other benefits that Intel cites are less relevant to our look at the -1280 v2, but better as you descend the product family. For instance, there are 11 total Xeon E3 models, allowing for broad differentiation. All but two models support Hyper-Threading, and all but two come equipped with 8 MB of shared L3 cache.

2. Intel’s Second-Gen Xeon E3 Processor Family

Aside from ECC memory support and four extra PCIe lanes, the Xeon E3s are very much similar to the third-gen Core processors we introduced in Intel Core i7-3770K Review: A Small Step Up For Ivy Bridge, architecturally.

There are more Xeon E3s, though. Intel already launched 11 different models with thermal ceilings as low as 17 W and as high as 87 W.

Xeon E3
Base Clock
Max. Turbo Boost
L3 Cache
Cores / Threads
HD Graphics
DDR3
Power
No Integrated Graphics
-1290 v2
3.7 GHz
4.1 GHz
8 MB
4 / 8
None
1600 / 1333
87 W
-1280 v2
3.6 GHz
4 GHz
8 MB4 / 8None1600 / 133369 W
-1270 v2
3.5 GHz
3.9 GHz
8 MB4 / 8None1600 / 133369 W
-1240 v2
3.4 GHz
3.8 GHz
8 MB4 / 8None1600 / 133369 W
-1230 v2
3.3 GHz
3.7 GHz
8 MB4 / 8None1600 / 133369 W
-1220 v2
3.1 GHz
3.5 GHz
8 MB4 / 4
None1600 / 133369 W
-1220L v2
2.3 GHz
3.5 GHz
3 MB
2 / 4
None1600 / 133317 W
Integrated Graphics
-1275 v2
3.5 GHz
3.9 GHz
8 MB4 / 8P4000
1600 / 133377 W
-1265L v2
2.5 GHz
3.5 GHz
8 MB4 / 82000
1600 / 133345 W
-1245 v2
3.4 GHz
3.8 GHz
8 MB4 / 8P4000
1600 / 133377 W
-1225 v2
3.2 GHz
3.6 GHz
6 MB
4 / 4
P4000
1600 / 133377 W


Three SKUs are rated at 77 W. Like the desktop chips bearing similar TDPs, these feature processor-based graphics (referred to as HD Graphics P4000). A fourth chip bears a lower 45 W ceiling, but runs at more conservative clock rates and sports HD Graphics 2000 instead.

The remaining seven Xeon E3s ship without graphics enabled, allowing Intel to scale back its thermal limits. Five of the products are rated at 69 W. A sixth gives up two of its cores to slip in at 17 W, and the seventh goes all-out with a 3.7 GHz base frequency, nudging power up to 87 W for the sake of performance.

What’s In A Name?

Given so many models differentiated in so many ways, it’s worth revisiting Intel’s nomenclature. Of course, we’re glad to see it using the same structure as last year. From my look at the Xeon E5s a few months back:

First, you have the brand, Xeon. Easy enough. Then there’s the product line: E3, E5, or E7. Again, we get the general sense that E3 is intended for entry-level single-socket workstations and servers, while E5 now spans a broader range from single- to quad-socket systems. The E7s cover two-, four-, and eight-socket servers.

The first digit you encounter specifies wayness, or the maximum number of CPUs in a node (that’s 1, 2, 4, or 8).

The second is indicative of socket type. Somewhat confusingly, Intel plans to use the numbers 2, 4, 6, and 8 moving forward. However, the actual interface corresponding to each digit may change. At least for 2012, we end up with the following associations:

2 = LGA 1155
4 = LGA 1356
6 = LGA 2011
8 = LGA 1567

The last two numbers are SKU designators like 10, 20, 30, and so on. Although there’s no formula to tell you why one chip might be a 50 and another a 70, Intel says it uses a combination of core count, cache size, clock rate, QPI data rates, and so on to classify each chip.

Certain models might also receive a single-letter suffix. For example, a model ending in L is meant as a low-power part. The CPUs we’re testing today are flagged as workstation models with a W suffix.

Finally, in the future, Intel plans to use a version number after the model name like v2 or v3 to identify generational progression. Ivy Bridge-based CPUs will be the first to employ those.

The time has come for those first v2-branded Ivy Bridge-based models, which simply succeed the first-gen parts. Also, two of the new Xeon E3s bear an L suffix, indicating their suitability in low-power (17 and 45 W) environments. Lastly, notice that the graphics-equipped chips all have names that end in a “5”, while the others end with a “0”.

Thus, it’s easy to interpret something like Xeon E3-1240 v2, a single-socket, LGA 1155-capable CPU roughly in the line-up’s middle. The “0” at the end tells us it doesn’t include processor graphics, and the lack of a suffix indicates standard voltage, making it a 69 W offering.

3. Platform Support: Three Old Chipsets, C216, And Memory Compatibility

Picking A Platform

As with Intel’s Ivy Bridge-based Core processors, the newest Xeon E3s are compatible with previous-generation motherboards, so long as vendors update their firmware to support them. Granted, upgrades are far less common in the server and workstation space. Technically, though, the C202, C204, and C206 chipsets work fine.

There is new core logic to complement Ivy Bridge-based Xeon E3s, though—C216.

You might recognize its Panther Point code name from Intel’s 7-series desktop chipsets, and its features largely from Z77. But C216 adds vPro and AMT 8.0 support, which are necessary for the remote management capabilities not available from Intel’s enthusiast-oriented offerings.


C216 (Workstation)
C206 (Workstation / Server)
C204 (Server)
C202 (Server)
vPro / AMT 8.0
X
X


Rapid Storage Technology
X
X
X
X
Smart Response Technology
X



Integrated Graphics
X
X


Supported Displays
Three
Two


HD Audio Support
X
X


Node Manager Support


X

USB 3.0 Ports
Four



USB 2.0 Ports
10
14
12
12
PCI Express 2.0 Lanes
Eight
EightEightEight
SATA 6Gb/s
Two
Two Two
SATA 3Gb/s
Four
FourFourSix


Otherwise, C216 facilitates DisplayPort 1.1 support, four USB 3.0 ports, HD Audio, a pair of SATA 6Gb/s ports (complementing four 3 Gb/s connectors), eight second-gen PCIe 2.0 lanes, and an integrated gigabit Ethernet MAC, just like Z77.

Like C206 before it, C216 is intended as a workstation-oriented chipset. It supports the Ivy Bridge architecture’s three display outputs on boards equipped with the right connectors. It has the audio and I/O functionality you’d expect to use on a desktop, but not a server. And it even adds Smart Response Technology to the company’s business portfolio, facilitating SSD-based caching for faster boot-up and application launching.

Memory Compatibility

The trickiest part of setting up our Xeon E3-1280 v2 and Intel S1200BTL motherboard was finding a memory kit that’d work. Like the desktop Core processors, these E3s support unbuffered modules-only. So, the 100 GB+ of registered modules we have on-hand don’t work.

Constrained to desktop-oriented kits, it quickly became clear that you want to pay close attention to Intel’s supported memory list prior to picking the pieces for a new server or workstation. We eventually tracked down four 2 GB modules based on Micron ICs, but not before exhausting four or five other kits from Kingston, G.Skill, and Crucial.

Platform
DIMM Configuration
Xeon E3-1200 v2 Family
Intel C202 and C204 Chipsets
Unbuffered, Non-ECC
Not Supported
Unbuffered, ECC
Supported
Unbuffered, Non-ECC/ECC Mix
Not Supported
Intel C216 and C206 Chipsets
Unbuffered, Non-ECCSupported (Client OS)
Not Supported (Server OS)
Unbuffered, ECCSupported
Unbuffered, Non-ECC/ECC MixNot Supported
Intel 7-Series Desktop Chipsets
Unbuffered, Non-ECCNot Supported
Unbuffered, ECCNot Supported
Unbuffered, Non-ECC/ECC MixNot Supported


The good news is that the E3’s memory controller is fairly flexible. It’s able to accommodate up to 32 GB in four slots, operating as fast as 1600 MT/s even with two DIMMS per channel. And although the C206 and C216 chipsets do support non-ECC mode, Intel’s configuration matrix clearly encourages you to stick with ECC-capable RAM.

4. Test Setup And Benchmarks
Test Hardware
Processors
Intel Xeon E3-1280 v2 (Ivy Bridge) 3.6 GHz, LGA 1155, 8 MB Shared L3, Hyper-Threading enabled, Power-savings enabled

Intel Xeon E3-1290 (Sandy Bridge) 3.6 GHz, LGA 1155, 8 MB Shared L3, Hyper-Threading enabled, Power-savings enabled

Intel Xeon E3-1275 (Sandy Bridge) 3.4 GHz, LGA 1155, 8 MB Shared L3, Hyper-Threading enabled, Power-savings enabled
Motherboards
Intel S1200BTL (LGA 1155) Intel C204, BIOS 35;1.13;1.14
Memory
Crucial 8 GB (4 x 2 GB) DDR3-1333 ECC Unbuffered, CT25672BA1339.18FG
Hard Drive
Intel SSD 710 200 GB SATA 3 Gb/s SSD
Graphics
Nvidia Quadro 5000
Power Supply
Enermax S Galaxy Evo EGX1250EWT 1250 W 80 PLUS Bronze PSU
System Software And Drivers
Operating System
Windows 7 Ultimate 64-bit
DirectX
DirectX 11
Graphics DriverNvidia Quadro Driver: 297.03

Although our focus is on workstation testing, we're using a more server-oriented motherboard based on Intel's C204 chipset, the S1200BTL. Really, this platform's only limitation is a PCI Express slot configuration designed for storage and networking expansion. As a case in point, the board's one 16-lane slot is wired for x8 data rates.

Also interesting was that, although we've seen desktop processors work in other vendors' workstation motherboards, the S1200BTL wouldn't recognize our Sandy or Ivy Bridge-based chips, limiting testing to the LGA 1155 Xeons in our lab.

Video Benchmarks and Settings
HandBrake CLIVersion: 0.9.5
Video: Big Buck Bunny (720x480, 23.972 frames) 5 Minutes, Audio: Dolby Digital, 48 000 Hz, Six-Channel, English, to Video: AVC Audio: AC3 Audio2: AAC (High Profile)
MainConcept Reference v2.2
Version: 2.2.0.5440
MPEG-2 to H.264, MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG-2), Audio:
MPEG-2 (44.1 kHz, 2 Channel, 16-Bit, 224 Kb/s), Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV
Application Benchmarks and Settings
WinRARVersion: 4.2
RAR, Syntax "winrar a -r -m3", Benchmark: 2010-THG-Workload
BlenderVersion: 2.62
Syntax blender -b thg.blend -f 1, Resolution: 1920x1080, Anti-Aliasing: 8x, Render: THG.blend frame 1, Cycles renderer and internal tile renderer (9x9)
e-on Software Vue 8 PLE
1920x1080 Landscape Render, Global Illumination enabled
Adobe Premiere Pro CS 5.5
Paladin Sequence to H.264 Blu-ray
Output 1920x1080, Maximum Quality, Mercury Playback Engine: Software Mode
Adobe After Effects CS 5.5
Version: CS 5.5
Tom's Hardware Workload, SD project with three picture-in-picture frames, source video at 720p, Render Multiple Frames Simultaneously
Adobe Photoshop CS 5.1 (64-Bit)Version: 11
Filtering a 16 MB TIF (15 000x7266), Filters:, Radial Blur (Amount: 10, Method: zoom, Quality: good) Shape Blur (Radius: 46 px; custom shape: Trademark sysmbol) Median (Radius: 1px) Polar Coordinates (Rectangular to Polar)
ABBYY FineReaderVersion: 10 Professional Build (10.0.102.82)
Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages
3ds Max 2012
Version: 10 x64
Rendering Space Flyby Mentalray (SPECapc_3dsmax9), Frame: 248, Resolution: 1440 x 1080
Cinebench
CPU Test, Built-in benchmark
SolidWorks 2010
PhotoView 360
Render 01-Lighter Explode.SLDASM (SolidMuse.com)
Image Output Resolution: 1920x1080, Render: Preview Quality “Good”, Final Render Quality “Best”
Visual Studio 2010
Compile Chrome project (1/31/2012) with devenv.com /build Release
5. Benchmark Results: Adobe CS 5.5

The Xeon E3-1280 v2 is the fastest processor in our Premiere Pro workload, but only by one second.

If you’ve seen this benchmark’s results in past workstation-oriented stories, then you know its outcome varies mostly based on the type of graphics card we use. Because we’re employing a Quadro 5000, CUDA support is triggered and the render takes one minute rather than nine or 10. Consequently, CPU performance is deemphasized, yielding the tight grouping between the three Xeon chips tested.

Although our Photoshop filters aren’t accelerated by CUDA, the numbers continue to place these three chips close together. The Ivy Bridge-based E3-1280 v2 does take first place by a slim margin, followed by the first-gen E3-1290 running at the same 3.6 GHz.

Adobe’s After Effects is baffling. Although it’s able to utilize many threads, it rarely seems to scale clearly based on processor performance. And the more threads you break an After Effects render into, the more memory the application seems to want. Because very specific compatibility requirements limited the number of modules in our lab that would work on Intel’s motherboard, it looks like After Effects probably would have been happier with more capacity.

6. Benchmark Results: Rendering

Our Vue 8 render workload takes more than 20 minutes, and Intel’s Ivy Bridge architecture helps cut more than 30 seconds from the task at the same 3.6 GHz clock rate. Shifting from Sandy to Ivy Bridge and adding 200 MHz nearly shaves off two minutes.

In testing Blender’s newer Cycles engine, we see slight scaling between last-gen and current-gen Xeon E3 CPUs. Not surprisingly, the -1275 finishes last due to its Sandy Bridge design and slower clock rate. The -1280 v2 places first again thanks to Intel’s Ivy Bridge architecture and 3.6 GHz base frequency.

Although we’d expect similar behavior from SolidWorks as Vue and Blender, the Ivy Bridge-based Xeon E3-1280 v2 establishes a larger advantage over the Sandy Bridge architecture in this one test.

Subtle improvements each step of the way make it pretty clear than upgrading a previous-generation workstation with a newer Xeon E3 probably won’t do much for performance (the same conclusion we drew about Intel’s desktop-oriented Core i7-3770K). However, the Xeon E3-1280 v2 is quantifiably faster than the Xeon E3-1290. So, if you held off on upgrading to a Sandy Bridge-based part, the second-gen E3s should yield more bang for your buck.

7. Benchmark Results: Transcoding

Known for its propensity to scale based on frequency, IPC throughput, and parallelism, MainConcept shows the biggest gap between Xeon E3 SKUs seen thus far. Absolutely, waiting six more seconds for a Xeon E3-1290 to finish its task isn’t major, but the 10% improvement could be significant in more demanding workloads.

Similar to the SolidWorks benchmark, HandBrake enjoys more of an advantage from Intel’s Ivy Bridge design that we see from the other benchmarks. Intel’s Xeon E3-1280 v2 achieves a strong win as a result.

8. Benchmark Results: Productivity

ABBYY’s OCR software is sensitive to clock rate, IPC, and the number of cores in a given architecture. Ivy Bridge doesn’t seem to help FineReader much, though, given the -1280 v2’s one-second advantage over the Xeon E3-1290 also running at 3.6 GHz.

There’s little difference between processors in Visual Studio 2010. Intel’s Ivy Bridge architecture maintains its first-place position, but not by much. Moreover, the faster (and more expensive) 3.6 GHz chips don’t see much benefit over the 3.4 GHz Xeon E3-1275 in this test.

Responding primarily to clock rate, the 3.6 GHz Xeon E3s (both the Sandy and Ivy Bridge-based parts) score similarly. They’re faster than the 3.4 GHz model, though.

9. Power Consumption

Rated for 69 W, the Xeon E3-1280 v2 is even more power-friendly than the Ivy Bridge-based Core i7-3770K.

The blue bar in the logged power consumption chart illustrates the Xeon E3-1280 v2 using less power, and the preceding pages proved that it delivers more performance in the process. In fact, on average, the -1280 v2-based system consumes 150 W with a Quadro 5000 skewing the power number higher.

In contrast, the slower last-generation Xeon E3-1290, rated for 95 W, drives the system averaging 161 W of power use. Meanwhile, the -1275-based configuration, which is slower still, sips 153 W, on average.

Yanking the Nvidia card would likely give Intel’s Xeon E3-1280 v2 a larger advantage, since the GPU adds a constant, higher draw. However, the -1280 v2 has no processor graphics, necessitating our platform upgrade.

10. Xeon E3-1200 v2 Is A Power Story, Not A Performance One

Based on our performance results, it’s clear that Intel’s Xeon E3-1280 v2 is a marginally-faster chip than its predecessors, the Xeon E3-1290 and -1275. The -1280 v2 operates at the same 3.6 GHz as the -1290, and stretches up to the same 4 GHz Turbo Boost limit. Both chips are quad-core parts with 8 MB of shared L3 cache and no processor graphics functionality.

In fact, it’s easier to list off the differences between them than their similarities.

There’s the Ivy Bridge architecture, most notably. The slight improvements Intel made to it confer, on average, a 3.8% speed-up at a given clock rate.

By simultaneously shifting to a 22 nm manufacturing process, Intel reduces the Xeon E3-1280 v2’s TDP by almost 30% compared to the 95 W Xeon E3-1290. In practice, across our benchmark suite, the -1280 v2 helped knock back power consumption by almost 7%. So, that’s almost 4% more performance at almost 7% less average platform power.

Now, compare where those two chips are selling. At launch, the Xeon E3-1290 was an $885 processor. The tray price on the -1280 v2 is $612—70% of the previous-generation flagship’s cost.

Upgrades are rare in the server and workstation space, so it’s highly unlikely that anyone would take a Xeon E3-1280 v2 and drop it into an existing LGA 1155-based box with an older CPU in it. However, system builders have to be happy about the prospect of a cheaper, faster, more power-efficient chip that works with existing motherboards.

I wasn’t particularly excited about Ivy Bridge in the desktop space. Meager performance improvements and disappointing overclocking made it impossible to recommend a new chip to anyone who purchased a new platform based on our endorsement of Intel’s Sandy Bridge design. But Xeon E3 targets a different customer. And the company’s entry-level server/workstation platform is more diverse than its desktop line-up based on the same design.

More than likely, integrators will shift over to second-gen Xeon E3s fairly transparently; you won’t have to make a conscious decision to seek out one of the new CPUs. But it’s good to know, once the performance data is collected and the power information compiled, that the newest LGA 1155-based Xeon chips do everything better than the models they replace, and often for a lot less money.