Page 1:Ivy Bridge Overclocking: What Does It Entail?
Page 2:Overclocking Ivy Bridge: Treating This Hot-Head Gingerly
Page 3:More Voltage, More Heat
Page 4:Digging Into Ivy Bridge's Overclocking Issues
Page 5:Practical Advice: Sandy Or Ivy Bridge?
Page 6:Test System And Benchmarks
Page 7:Benchmark Results: Professional Applications
Page 8:Benchmark Results: Adobe CS 5.5
Page 9:Benchmark Results: Audio/Video
Page 10:Benchmark Results: Matlab
Page 11:Benchmark Results: File Compression And Power Consumption
Page 12:Single- And Multi-Threaded Efficiency
Page 13:Overall Efficiency
Page 14:Ivy Bridge Takes A Bronze In Overclocking; Gold In Efficiency
More Voltage, More Heat
Our evidence shows that, at similar clock rates, 32 nm Sandy Bridge-based processors don't get as hot as the new 22 nm Ivy Bridge-based chips. As a result, our Core i7-3770K dropped its overclocked frequency sooner in order to protect the CPU's integrity, nullifying the effects of our overclocking efforts entirely. It also discourages setting higher core voltages, since that'd just exacerbate the issue.
Thus, throttling can only be avoided when the target overclock is achieved using the lowest core voltage possible. However, it still has to be high enough to allow the transistors to operate reliably. Old mantras like “a hefty voltage increase helps facilitate a hefty frequency bump” do not work on Ivy Bridge-based chips. Once throttling kicks in, average clock rates drop to around 3.6 GHz. That's a significant reduction compared to the 4.6 GHz we were targeting.
We think that you're going to have the most success overclocking Ivy Bridge if you can stay as close as possible to the CPU's default voltage setting or work really hard to overcome the heat resulting from more aggressive voltage settings. At 4.6 GHz, the voltage increase necessary to get our machine stable quickly resulted in throttling, forcing us to stay at 4.5 GHz. At that speed, we had to increase the core voltage by 70 mV in order for our benchmark suite to run reliably.
As in our launch coverage, 4.5 GHz turned out to be as high as we could go without running into stability issues. It wasn't difficult to get up to 4.7 GHz (or, using one of our samples, even 4.9 GHz). But certain tests consistently cause those frequencies to derail, pushing us back down.
- Ivy Bridge Overclocking: What Does It Entail?
- Overclocking Ivy Bridge: Treating This Hot-Head Gingerly
- More Voltage, More Heat
- Digging Into Ivy Bridge's Overclocking Issues
- Practical Advice: Sandy Or Ivy Bridge?
- Test System And Benchmarks
- Benchmark Results: Professional Applications
- Benchmark Results: Adobe CS 5.5
- Benchmark Results: Audio/Video
- Benchmark Results: Matlab
- Benchmark Results: File Compression And Power Consumption
- Single- And Multi-Threaded Efficiency
- Overall Efficiency
- Ivy Bridge Takes A Bronze In Overclocking; Gold In Efficiency