Test Methodology
Intel Core Processor Configurations |
|||||||
---|---|---|---|---|---|---|---|
CPU | |||||||
CPU Base Clock | 3.7GHz |
4.1GHz |
3.6GHz |
3.6GHz |
3.7GHz |
2.9GHz |
3.6GHz |
CPU Turbo Clock | 5.2GHz |
4.8GHz |
5.0GHz |
4.9GHz |
4.6GHz |
4.1GHz |
4.2GHz |
CPU Cache | 20MB |
12MB |
16MB |
12MB |
9MB |
9MB |
6MB |
CPU Cores / Threads | 10 / 20 |
6 / 12 |
8 / 16 |
8 / 8 |
6 / 6 |
6 / 6 |
4 / 4 |
CPU TDP | 125W |
125W |
95W |
95W |
65W |
65W |
65W |
Integrated Graphics | UHD 630 |
UHD 630 |
UHD 630 |
UHD 630 |
UHD 630 |
- |
- |
IGP Base Clock | 350MHz |
350MHz |
350MHz |
350MHz |
350MHz |
- |
- |
IGP Turbo Clock | 1.2GHz |
1.2GHz |
1.20GHz |
1.20GHz |
1.15GHz |
- |
- |
Socket | LGA 1200 |
LGA 1151v2 |
|||||
Lithography | 14nm |
||||||
Motherboard | MSI MPG Z490 Gaming Carbon WiFi |
Asus ROG Strix Z390-I Gaming |
|||||
BIOS | E7C73IMS.123 |
2603 |
|||||
Memory | G.Skill Trident Z DDR4-3200 |
||||||
Memory Config | 2x16GB |
||||||
Memory Timings | 14-14-14-34-1T |
||||||
Memory Speed | 2,933MHz |
2,666MHz |
2,666MHz |
2,666MHz |
2,666MHz |
2,666MHz |
2,400MHz |
Disk Drive | Corsair MP600 NVMe 1TB |
||||||
Power Supply | be quiet Dark Power Pro 11 (1,000W) |
||||||
Graphics Card | Nvidia GeForce RTX 2080 Ti Founders Edition (445.87) |
||||||
CPU Cooler | Noctua NH-D15S |
||||||
Operating System | Microsoft Windows 10 (64-bit) 1909 |
AMD Ryzen Processor Configurations |
||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
CPU | ||||||||||||||||
CPU Base Clock | 3.5GHz |
3.GHz |
3.9GHz |
3.8GHz |
3.8GHz |
3.8GHz |
3.6GHz |
|||||||||
CPU Turbo Clock | 4.7GHz |
4.6GHz |
4.5GHz |
4.4GHz |
4.4GHz |
4.3GHz |
3.9GHz |
|||||||||
CPU L3 Cache | 64MB |
64MB |
32MB |
32MB |
32MB |
16MB |
16MB |
|||||||||
CPU Cores / Threads | 16 / 32 |
12 / 24 |
8 / 16 |
8 / 16 |
6 / 12 |
4 / 8 |
4 / 8 |
|||||||||
CPU TDP | 105W |
105W |
105W |
65W |
95W |
65W |
65W |
|||||||||
Integrated Graphics | - |
- |
- |
- |
- |
- |
- |
|||||||||
IGP Base Clock | - |
- |
- |
- |
- |
- |
- |
|||||||||
IGP Turbo Clock | - |
- |
- |
- |
- |
- |
- |
|||||||||
Socket | AM4 |
|||||||||||||||
Lithography | 7nm |
|||||||||||||||
Motherboard | Asus X570 Crosshair VIII Hero WiFi |
|||||||||||||||
BIOS | 1302 |
|||||||||||||||
Memory | G.Skill Trident Z DDR4-3200 |
|||||||||||||||
Memory Config | 2x16GB |
|||||||||||||||
Memory Timings | 14-14-14-34-1T |
|||||||||||||||
Memory Speed | 3,200MHz |
|||||||||||||||
Disk Drive | Corsair MP600 NVMe 1TB |
|||||||||||||||
Power Supply | be quiet! Dark Power Pro 11 (1,000W) |
|||||||||||||||
Graphics Card | Nvidia GeForce RTX 2080 Ti Founders Edition (445.87) |
|||||||||||||||
CPU Cooler | Noctua NH-D15S |
|||||||||||||||
Operating System | Microsoft Windows 10 (64-bit) 1909 |
Benchmark Suite |
|
---|---|
CPU | |
HEXUS PiFast | Our number-crunching benchmark tests the single-thread power of the CPU. |
Cinebench Release 20 | This popular cross-platform benchmark examines both single- and multi-core performance. |
Blender 2.82 | We render the Classroom scene in this popular open-source creation suite. |
Memory and Storage | |
AIDA64 | Benchmark that accurately depicts memory bandwidth and latency. |
System | |
PCMark 10 | A set of tests covering a wide variety of tasks performed in the modern workplace. |
VRMark | Using the DirectX 12 API, the Cyan Room test examines a system's VR credentials. |
3DMark | We run Time Spy, a QHD DirectX 12 test of both GPU performance and stability. |
Gaming | |
Civilization VI: Gathering Storm | Tested in DX11, with 4x MSAA and Ultra quality settings. |
Far Cry: New Dawn | Tested in DX11, with TAA, Ultra quality settings and HD textured enabled. |
Final Fantasy XIV: Shadowbringers | Tested in DX11, with FXAA and Maximum quality settings. |
Shadow of the Tomb Raider | Tested in DX12, with TAA and Highest quality settings. |
Vitals | |
CPU Temperature | Core temperature recorded when idle and when rendering via Blender. |
Power Consumption | System-wide mains power draw recorded when idle and when playing Shadow of the Tomb Raider. |
Notes
It's worth mentioning there were multiple motherboard BIOSes released in the run-up to launch. The primary issue was one of ensuring the chips turbo boosted correctly when BIOSes were set to Auto. As shown on the first page, Intel says the Core i9-10900K can run at up to 5.3GHz on a couple of cores and, temperature permitting, 4.9GHz across all cores.
Achieving such speeds requires tweaking the power levels of the processor. Intel chips are equipped with at least a couple of exposed power levels known as PL1 and PL2. The latter is the short-term limit that is higher than the TDP, which in this case is 125W. PL1 is the long-term power, which ought to be that 125W, and the time the processor stays in the short-term PL2 is known as Tau.
Now, the processor maintains its 4.8GHz/4.9GHz (TVB) all-core speed when it's in the PL2 state, but according to Intel's guidelines, drops down to something lower when PL1 is active. Benchmarks that are over quickly, within the PL2 envelope, do well, but those that are longer ought to spill over to lower-frequency PL1.
Intel, however, offers system integrators and motherboard partners a choice on how they implement these power targets and Tau time. "We have expanded our definition of PL2 to allow customers to take further advantage of a 10th Gen Intel Core desktop processor’s performance options and capabilities. Our customers now have more ability to configure their boards and implement power and performance settings according to their own design targets. Some customers may want to follow Intel’s configuration recommendation, while others will design boards that draw more power and may require more capable cooling solutions".
All of this is important because the motherboard guys want to hit peak performance for as long as possible. The flurry of recent BIOSes aim to achieve that peak 5.3GHz TVB frequency and 4.9GHz all-core load for lengthy benchmarks. Even up until last Friday BIOSes were being released by all the major makers, including MSI, as tested here, to fully exploit the potential speed of the processor.
The upshot of all this is that, set to Auto in BIOS, the Core i9-10900K easily surpasses its 125W TDP for long-term load, and operates at somewhere in the region of 200W, unlike AMD processors that steadfastedly keep within their allocated TDP. You'll see exactly what we mean on the power-consumption page.