Intel Core i7-11700 and 11700K challenge AMD Ryzen 7 5800X. The price advantage on their side, but they bite their teeth, because they offer not only the strengths of Rocket Lake-S and eight cores, but also the weaknesses.
Table of contents
- 1 Intel Core i7-11700 (K) against AMD Ryzen 7 5800X
- Overview of the eight-core from Rocket Lake-S
- TDP and consumption at a glance
- Intel Core i7-11700: 224 watts for 28 seconds – theoretically
- EWMA and the BIOS decide
- 2 benchmarks in applications and games
- How ComputerBase tests
- Tests in multi-core applications
- Tests in single-core applications
- Tests in games in three resolutions
- 3 Power consumption, efficiency, temperature and overclocking
- Power consumption and temperature in applications
- Power consumption in games
- CPU overclocking almost impossible, but RAM very simple
- 4 Conclusion and recommendation
Intel's current desktop processor family Rocket Lake is available as a variant with eight CPU cores not only for the flagship Core i9-11900K for almost 600 euros, but also for the Core i7. The versions Core i7-11700F and 11700 stand out, which cost almost half at 310 and 325 euros, respectively. But the Core i7-11700K with almost twice as high a TDP can be an option that is hardly more expensive.
Of course, certain compromises have to be accepted for this. The test of the Core i7-11700 and 11700K as variants that cover all regular models of this series will show how big they turn out, what is ultimately available for a little more than half the money for performance and how they are positioned compared to AMD's strong Ryzen 5000 , show.
The two test models, the Intel Core i7-11700K and the Intel Core i7-11700, were available as a boxed version directly from the trade from the Caseking * online shop for Provided.
Overview of the eight-core from Rocket Lake-S
There are ten models from Intel in the Rocket Lake-S family with eight cores, which can be seen as exaggerated. The difference lies in the detail, but it is precisely this detail that is sometimes so small that it hardly justifies an additional model – the smaller versions therefore come off particularly well here. Because the surcharge for a Core i9 compared to a Core i7 is more than obvious in some areas. It amounts to a maximum of 145 US dollars, as shown in the example 11700KF to 11900KF.
Larger differences in Intel prices
As the table reveals, the list price is one thing, the real retail price is quite another. Here the solutions not only separate 120 to 140 US dollars, but significantly more. The now massive price difference for at first glance only “a few MHz” is definitely too much for some variants – the small bonuses don't help either. Because while the 11900K is the only DDR4-3200 with Gear 1 that is still appealing, this is not an official part of the specification for the small 11900s – but the price of the 11900F is much better. This also applies to the Adaptive Boost, which, as explained in the original article, unpacks the crowbar as a turbo for all cores. The 11700 models don't have any of that.
If the CPUs are operated with the official specifications, the air for all extra modes is very thin anyway. Then it's really just a question of binning and the aforementioned difference in the maximum possible clock rate for one or more cores. But this applies primarily to applications, in games it is even more similar.
The opponent: AMD Ryzen 7 5800X
Little needs to be said today about the opponent of the eight-core from Intel. The AMD Ryzen 7 5800X is now easily available from EUR 374 (best price *) well below the RRP of EUR 449. The model brings its 8 Zen 3 cores and 16 threads to a clock rate of 4.7 GHz with a TDP of 105 watts. It has become a recommendation, especially because of its gaming strength, but it doesn't have to be afraid of Intel when it comes to applications either, as the course of the test once again shows.
TDP and consumption at a glance
A keyword for all new Intel processors are the TDP and consumption. At first glance, Intel was surprisingly open to announcing the new CPUs on this point. It was said that Rocket Lake-S would by and large take over the TDP framework from Comet Lake-S and this statement was underpinned with a lot of information on consumption and current strengths.
At second glance, however, it became clear that some information was missing. They show that Intel has again expanded the framework in which the CPUs are allowed to operate according to the official specification for some models.
K-models with up to 35% more PL2
The top model is not affected, but the two smaller K CPUs are. The Core i5-11600K is now allowed to call up up to 250 watts for a maximum of 56 seconds (PL2) according to the specification, previously the maximum was set at 184 watts – an increase of 35 percent. And for the Core i7-11700K, the official consumption limit rises from 229 to 250 watts.
Tau increased from 28 optionally to 56 seconds
Also new is the tau, i.e. the Time that a CPU can call up a maximum of PL2, optionally also with smaller processors, can be 56 seconds. Before that, it always ended at 28 seconds. But the information is more of a theoretical nature, as later revealed in the test.
The question of the long-term power consumption of the 65-watt model is also not ignored in the test. Mainboards for hobbyists like to let Intel's CPUs run free and only put on shackles if this is explicitly desired. In the OEM environment, however, it can quickly turn out the other way around and the limit values for PL1, PL2 and Tau specified by Intel are inevitable. Since the smallest Core i7 is a hot candidate for both markets, the editors are also looking at operation with official limits in the test.
With AVX -512 only slightly above the base clock
The brutal thing is again the loss of clock speed when everything is demanded by the CPU and the latest instructions take effect: read out using a log file in continuous operation, the clock speed of 65 watts is quickly only 3.1 GHz, with full AVX load even completely well below the 3 GHz mark at an average of 2.7 GHz. The base clock rate of 2.5 GHz for the Core i7-11700 defined by Intel is not so low for nothing. Without the provided peak loads and turbo mode, the processor would be completely different today. If you let go, clock rates of 4.4 GHz are the norm, even when all cores are under load.
«Previous CPU cycle (blender, multi-core) CPU package power (blender, multi-core) CPU temperature (blender, multi-core) Next»
- CPU clock (blender, multi-core) 01.0002.0003.0004.0005.000MHz 159131721252933374145495357616569737780Seconds Intel Core i7-11700 – with limits (65 watts after max. 28 seconds) Intel Core i7-11700 – without limits
- CPU package power (blender, multi-core) 04080120160200Watt (W) 14710131619222528313437404346495255586164677073767980Seconds Intel Core i7-11700 – with limits (65 watts) 28 seconds) Intel Core i7-11700 – no limits
- CPU temperature (blender, multi-core) 020 406 080 ° C 14710131619222528313437404346495255586164677073767980 seconds Intel Core i7-11700 – with limits (65 watts after max. 28 seconds) Intel Core i7-11700 – without limits
The Core i7-11700K reaches an AVX clock of 3.7 GHz under 125 watt continuous load. Here, too, the limitation is clear, because Intel outputs 3.6 GHz as the base clock. After testing the fifth Rocket Lake CPU, we can clearly see that the processor only operates around 100 MHz above the base clock when the AVX is heavily loaded. The value, which the editors used to call “almost never applied basic clock”, has again received significantly more relevance to reality.
This again confirms what ComputerBase already reported for the first article: Rocket Lake not only buys more performance with IPC, but above all with increased power consumption. If the shackles are put on hard, it can become extremely tight for the new architecture in some scenarios – not only at the top, but also at the lower end of the service.
EWMA and the BIOS decide
How quickly a Core i7-11700F operated according to specifications is throttled to 65 watts is also decided in this case by the EWMA – the weighted moving average of the power consumption. 28 seconds is more of a theoretical maximum value, because under peak load the average power consumption reaches PL1 much faster, which is why the CPU – operated according to official specs – can fall back from PL2 to PL1 even earlier than after 28 seconds.
This can be seen very clearly in Cinebench and Blender: After around 15 seconds, the EWMA budget for the 11700F, which was always working with well over 150 watts of power, is already exhausted and the processor has to shut down to 65 watts. The clock drops significantly. On the plus side, however, is an extremely low CPU temperature, which now remains below the 50 ° C mark.
«Previous 3 x CB R20, 2 sec. Break (Package Power) 3 x CB R20, 2 sec . Pause (clock) 3 x CB R20, 2 sec. Pause (temperature) Next »
- 3 x CB R20, 2 sec. Pause (Package Power) 04080120160200Watt (W) 110192837465564738291100109118127136145154163172181190199208217226230 Intel Core i7-11700 – with limits (65 Watt after max. 28 seconds) Intel Core i7-11700 – without limits.
- 3 x CB R20, 2 sec. Break (temperature) 020 406 080 ° C 110192837465564738291100109118127136145154163172181190199208217226230 Intel Core i7-11700 – with limits (65 watts after max. 28 seconds) Intel Core i7-11700 – without limits
= “js-chart-carousel-li nojs-block” role = “presentation” hidden> 3 x CB R20, 2 sec. pause (clock) 01.0002.0003.0004.0005.000MHz 110192837465564738291100109118127136145154163172181190199208217226230 Intel Core i7-11700 – 65 after max. 28 seconds) Intel Core i7-11700 – no limits
< p class = "p text-width">The reason: PL1 may only be exceeded up to PL2 as long as the moving average (EWMA, Exponentially Weighted Moving Average) of the CPU power consumption has not already reached the level PL1 or, alternatively, 28 seconds have passed. In other words: In the time-weighted average, an Intel processor never consumes more than the TDP according to the specifications. Since the very high peak value causes the moving average to skyrocket very quickly, 28 seconds are never used in such scenarios. The optional 56 seconds rope that Intel offers are therefore not implemented by most board manufacturers.
Another CPU with 4,096 watts
If the mainboard ignores Intel's consumption specifications PL1, PL2 and Tau per se and thus always lets the CPU run at the maximum allowed turbo clock, nothing changes. In particular in OEM PCs that implement Intel's specifications in order to get by with a cooling system that meets the TDP, higher consumption in the short term means a gain in performance with short loads. This is exactly how the modes are called by some manufacturers: MSI uses the exact Intel specifications for the “Boxed cooler” setting and a mixture of higher and sometimes very high PL1 limits for “High-End Cooler”. With the “water cooling” setting, the manufacturer simply lets the CPU run completely without any limitation, which the value 4,096 watts simply stands for. It is obvious that the performance is different.
(*) The links marked with an asterisk are affiliate links. In the case of an order via such a link, ComputerBase receives a share of the sales proceeds without increasing the price for the customer.
On the next page: Benchmarks in applications and games