As announced for the 2nd quarter at CES 2021 in January, Intel has now launched the market for the first 10 nm notebook CPUs with up to 8 modern Willow Cove Cores and Xe-Graphic aka “Tiger Lake-H45” Posted. One million CPUs, including the top model Core i9-11980HK, are already being shipped by Intel.
Tiger Lake H (45): Up to 8 Cores with up to 107 watts
While the Tiger Lake-H35 (test) CPUs presented at CES in January with a maximum TDP of 35 watts were only slightly higher clocked Tiger Lake-U with still only four cores and four PCIe lanes according to Standard 4.0 thanks to higher consumption, Tiger uses Lake-H45 now has a new die that offers 8 cores or 6 cores and 20 PCIe 4.0 lanes.
-
Image 1 of 4
< p class = "p text-width">Compared to the predecessors of the Comet Lake-H type, the generation change also brings an average IPC gain of almost 20 percent in this case, but the clock rate is slightly lower. id = “section_ipcgewinn_vs_taktrueckgang”> IPC profit vs. clock rate decrease
As with the Tiger Lake H35 top model with Turbo 3.0, the maximum clock rate remains at 5.0 GHz and, in the overclockable top model, the Core i9-11980HK and the two largest professional offshoots, the W-11955M and i9-11950H, it shouldn't just be on one, but rather can be applied to up to two cores at the same time. Comet Lake-H was able to clock up to 5.3 GHz, which is still not reliably possible with the 10 nm process. That is why Intel had to port the new architecture for the new Rocket Lake S desktop CPUs to 14 nm.
Intel gives the highest clock when under load on all eight cores with 4.5 GHz, the consumption should then be up to 107 watts (PL2) like the HK predecessor and should only be present for a limited time – if at all – in most notebooks.
Intel continues to soften the TDP classes
At the start of the new CPUs, Intel no longer speaks of Tiger Lake-H45 or Tiger Lake-H35, but rather of Tiger Lake-H. This makes sense in view of the development of notebook CPUs over the past few years, because the clear classes (15 watt TDP for U or 45 watt TDP for H) have long ceased to exist. Intel is now giving OEMs significantly more leeway in the choice of PL1 (long-term consumption maximum, TDP) and PL2 (short-term consumption maximum), anchoring CPUs of a series by name to a fixed TDP hardly corresponds to reality.
What Intel is now communicating instead is the maximum possible 1, 2, 4, 6 and 8 core clock (5.0 GHz should be possible on one core at around 35 watts, 4.5 GHz on eight Cores at approx. 107 watts) and the typical all-core turbo at reduced TDP (cTDP), which is 65 watts for the overclockable i9-11980HK and 35 watts for the other models. The Core i9-11980HK is supposed to deliver 3.3 GHz at 65 watts with load on all cores, with PL2 = 107 watts it should be up to 4.5 GHz.
threads max. All-core clock
min./max. TDP max. Clock
(number of cores) L3 cache graphic EUs max. GPU clock DDR4 TDP Tiger Lake H45 with 6 or 8 cores (new) Core i9-11980HK 8/16 3.3 GHz (65 W)/4.5 GHz 5.0 GHz (2 core) 24 MB UHD 32 1.45 GHz 3200 45-65 Watt Core i9-11900H 8/16 2, 1 GHz (35 W)/4.4 GHz 4.9 GHz (2 core) 24 MB UHD 32 1.45 GHz 3200 35-45 Watt Core i7-11800H 8/16 1.9 GHz (35 W)/4, 2 GHz 4.6 GHz (2 cores) 24 MB UHD 32 1.45 GHz 3200 35-45 watts Core i5-11400H 6/12 2.2 GHz (35 W)/4.1 GHz 4.5 GHz (2 cores ) 12 MB UHD 32 1.45 GHz 3200 35-45 Watt Core i5-11260H 6/12 2.1 GHz (35 W)/4.0 GHz 4.4 GHz (2 core) 12 MB UHD 32 1.40 GHz 3200 35-45 watts Tiger Lake H45 with 6 or 8 cores and vPro (new) Xeon W-11955M 8/16 2.1 GHz (35 W)/4.5 GHz 5.0 GHz (2 core) 24 MB UHD 32 1.45 GHz 3200 (ECC) 35-45 Watt Core i9-11950H 8/16 2.1 GHz (35 W)/4.4 GHz 4.9 GHz (2 core) 24 MB UHD 32 1.45 GHz 3200 35-45 watt Xeon W-11855M 6/12 2.6 GHz (35 W)/4.4 GHz 4.9 GHz (2 core) 18 MB UHD 32 1.45 GHz 3200 (ECC) 35-45 Watt Core i7-11850H 8/16 2.1 GHz (35 W)/4.3 GHz 4 , 8 GHz (2 core) 24 MB UHD 32 1.45 GHz 3200 35-45 Watt Core i5-11500H 6/12 2.4 GHz (35 W)/4.2 GHz 4.6 GHz (2 core) 12 MB UHD 32 1.40 GHz 3200 35-45 watts Tiger Lake H35 with 4 cores Core i7-11375H SE 4/8 3.0 GHz (28 W)/3.3 GHz 5.0 GHz (1 core) 12 MB Iris Xe 96 1.35 GHz 3200/LP-4266 28-35 Watt Core i7- 11370H 4/8 3.0 GHz (28 W)/3.3 GHz 4.8 GHz (1 core) 12 MB Iris Xe 96 1.35 GHz 3200/LP-4266 28-35 Watt Core i7-11300H 4/8 2.6 GHz (28 W)/3.1 GHz 4.4 GHz (1 core) 8 MB Iris Xe 80 1.30 GHz 3200/LP-4266 28-35 watts For comparison: the fastest U-series CPU core i7-1185G7 4/8 1.2 GHz (12 W)/3.0 GHz 4.8 GHz (1 core) 12 MB Iris Xe 96 1.35 GHz 3200/LP-4266 12-28 watts
< h2 class = "text-width text-h2" id = "section_intel_xe_mit_maximal_32_statt_96_eus"> Intel Xe with a maximum of 32 instead of 96 EUs
Intel's H series is intended for notebooks with a dedicated GPU. Because the Xe graphics unit is usually only used in 2D mode and for video encoding (Quick Sync), Intel has it compared to Tiger Lake-U/Tiger Lake-H35 from a maximum of 96 (G7) to only 32 EUs reduced. This frees up space on the wafer for more important components. The architecture is still Xe, but the small iGPU can only be called like its predecessor: Intel UHD Graphics.
20 PCIe lanes after Standard 4.0 and AVX512
The new CPUs offer 20 PCIe lanes according to Standard 4.0, which can be used to connect memory or graphics cards. Eight more are used for communication with the chipset. Four of the 20 that can be used by OEMs are always intended for an NVMe slot. The remaining 16 can be configured as 2 × 8 or 1 × 8 and 2 × 4. The chipset also provides 24 additional PCIe lanes according to Standard 3.0.
For the first time “professional CPUs” in the H series
“Professional” CPUs are celebrating a premiere in the H series. Instead of giving some SKUs functions like vPro as in the past and not others, there are now explicitly models with vPro, the name of which differs from the consumer models. This has been the case with U-CPUs for years. As with the predecessor, two of the new models will be launched as Xeon W and not as Core i. The Xeon W-11955M corresponds to the Core i9-11980HK expanded by the Pro functions without a free multiplier, while the Xeon W-11855M has no consumer counterpart with 18 MB L3 cache and only 6 cores. Xeon W also exclusively supports memory with ECC.
Intel wants to clearly beat AMD in games
As always, manufacturer benchmarks should be treated with great caution. They are by no means wrong, but extremely well selected and often only show the very best. Most recently, the game selection in particular had shown that you can turn a result completely in favor of a manufacturer with just a few titles, so a double-digit lead can turn into a double-digit deficit if desired. This of course makes comparison tests all the more difficult, because the accusation can always be made that one prefers manufacturer A or B.
The picture drawn by Intel for the market launch is definitely clear: The top model Core i9-11980HK should clearly beat AMD's top model Ryzen 9 5900HX and the 6-core processor Core i5-11400H still beat the Ryzen 9 5900HS with a reduced TDP. And that although the Nvidia GeForce RTX 3000 laptop GPU used in the respective notebooks in the model with AMD Ryzen can always draw a little more power.
-
Image 1 of 2
Market start today, sales start (much) later
Intel claims to have shipped over a million of the new processors to OEMs to date. The first products on the market can be expected in the coming week at the earliest, only towards the end of the 2nd quarter and thus at the time of the – canceled – Computex, more and more of the 80 partner notebooks planned with Tiger Lake-H45 will come onto the market . In addition, no OEM has so far been able to offer the editorial team a test sample.
15 of 80 should come with vPro CPUs, none of them will carry the Evo certificate. This is not surprising, however, as Evo is only intended for notebooks without dGPU.
-
Image 1 of 3
ComputerBase received information about this article from Intel under NDA. The only requirement was the earliest possible publication time.