Over the last few months the 20-threaded Xeon E5-2660 v3 processor has gained considerable popularity among thrifty gamers: it’s cheap, has a very considerable Turbo Boost frequency, and in addition, boasts an impressive amount of L3 cache. And all this is priced at a mere $23.

In this article we will try to find out if the 2660 v3 will be able to be worthy competition to cheaper actual chips like E5-2620 v3 and E5-2630 v3.

CPU

Here is a 10-core, 20-thread processor based on the relatively current Intel Haswell-EP architecture, codenamed Xeon E5-2660 v3. Let’s take a look at the chip itself:

We got a sample marked SR1XR, the chip is made in Malaysia, its rated frequency is quite small 2600MHz, and TDP is set at 105 watts. In general, such processor will be able to cool actually any cooler with three or four heat pipes.

At the heart of the E5-2660 v3 is a 22nm, 10-core, 20-threaded variation of the Haswell-EP R2 revision die. This chip is designed for installation in LGA 2011 v3 socket motherboards, its nominal frequency is 2600MHz, but with Turbo Boost 2.0 technology it can increase significantly: 3300MHz for a single core and 2900MHz for all 10 cores, 20 threads. But this is easy to fix.

At the moment there are an infinite number of modifications of LGA 2011 v3 motherboard bios, which allow to make all processor cores/threads function at maximum Turbo Boost clock speed for one core. Thus, our test E5-2660 v3 can quite stably operate at 3300MHz for all 10 cores, 20 threads. Which we took advantage of:

The undervoltage settings for our E5-2660 v3: -60mv, -50mv, -50mv.

Other features of Xeon E5-2660 v3 are as follows: Chip has 256KB of L2 cache per core and a total, really big for its time, L3 cache of 25MB. CPU built in memory controller speed is fixed at 3000MHz, L3 cache functions at the same speed. Maximum RAM frequency is limited by four-channel DDR4-2133MHz controller.

Besides all the features mentioned above, it should be noted that the 10-core Haswell has SSE4.2, AVX and AVX2 instructions, which are so necessary in the realities of 2022. This feature will help hardcore gamers to run practically any novelty of this year without any problems, and, as practice shows, even with activated ray tracing.

The E5-2660 v3 performance test will be performed only in Turbo Boost “unlocked” mode at 3300MHz.

Test setup, software and CPU settings

Test setup

  • Intel CPUs — Xeon X3470, Core i7-2600, Xeon X5660, Xeon E5-2667, Xeon E5-2620 v3, Xeon E5-2630 v3, Xeon E5-2660 v3;
  • AMD CPUs — FX-8350, FX-6350;
  • CPUs cooling — Cooler Master Hyper 212 Black Edition (RR-212S-20PK-R1);
  • LGA 1155/1156 RAM — 4 x 8GB HyperX Genesis Na’Vi Edition (KHX16C9C2K2/8);
  • LGA 1366 RAM — 3 x 8GB HyperX Genesis Na’Vi Edition (KHX16C9C2K2/8);
  • LGA 2011 RAM — 4 x 8GB Micron MT36JSF1G72PZ-1G4M1HF;
  • LGA 2011 v3 RAM — 4 x 4GB G.SKILL DDR4 F4-2400C15S-4GNT;
  • AM3+ RAM — 4 x 8GB HyperX Genesis Na’Vi Edition (KHX16C9C2K2/8);
  • LGA 1156 Motherboard — DELL OptiPlex 980 (0D441T, mod-BIOS);
  • LGA 1155 Motherboard — Gigabyte GA-Z68P-DS3 (rev. 2.0);
  • LGA 1366 Motherboard — DELL T3500 (09KPNV);
  • LGA 2011 Motherboard — DELL T3610 (09M8Y8);
  • LGA 2011 v3 Motherboard — Qiyida X99-H9 with modified BIOS from Huananzhi x99-8m-f (Unlock Turbo Boost, as well as unlocked CPU timings and voltage control);
  • AM3+ Motherboard — ASUS M5A97 LE R2.0;
  • Videocard — KFA2 GeForce RTX 2060 SUPER 8GB (~1950/14816MHz, Power Limit 112%);
  • SSD — KINGSTON SUV400S37120G 120,0 GB (Windows 10), KINGSTON SUV400S37240G 240,0 GB (Games), KINGSTON SUV400S37400G 400,0 GB (Games);
  • PSU — Chieftec GPS-1250C.

Software

  • Operating system: Windows 10 x64 with the latest updates as of September 2022;
  • Videocard drivers: NVIDIA GeForce 516.94 WHQL;
  • Additional FPS measuring software: MSI Afterburner 4.6.4;
  • Games: Testing was conducted on actual game versions as of September 2022;
  • Game settings: testing was performed at the highest possible graphics settings at 720p resolution.

Setting up the processors under test

  • Intel CPUs:
    • Core i7-2600@3400MHz, Dual-Channel DDR3@1866MHz (DDR3 timings 10-11-11-26, DDR3 voltage 1.550v);
    • Xeon X3470@3200MHz, Dual-Channel DDR3@1333MHz;
    • Xeon X5660@3066MHz, Triple-Channel DDR3@1333MHz;
    • Xeon E5-2667@3200MHz, Quad-Channel DDR3@1333MHz;
    • Xeon E5-2620 v3@3200MHz, Quad-Channel DDR4@1866MHz (DDR4 timings 10-10-10-24);
    • Xeon E5-2630 v3@3200MHz, Quad-Channel DDR4@1866MHz (DDR4 timings 10-10-10-24);
    • Xeon E5-2660 v3@3300MHz, Quad-Channel DDR4@2133MHz (DDR4 timings 12-12-12-32);
  • AMD CPUs:
    • FX-6350@4200MHz, CPU-NB@2200MHz, Dual-Channel DDR3@1866МГц;

    • FX-6350@4595MHz, CPU-NB@2608MHz, Dual-Channel DDR3@2086MHz (CPU Core voltage — 1.440v, CPU-NB voltage — 1.375v, DDR3 voltage — 1.660v);

    • FX-8350@4200MHz, CPU-NB@2200MHz, Dual-Channel DDR3@1866MHz;

    • FX-8350@4434MHz, CPU-NB@2608MHz, Dual-Channel DDR3@2086MHz (DDR3 timings 11-12-12-30, CPU Core voltage — 1.416v, CPU-NB voltage — 1.375v, DDR3 voltage — 1.660v).

You can find out more about almost all tested processors from our full-fledged materials, or by checking out our blogs section, where small articles are posted.

Separately, let us remind the methodology of CPU testing: each game was run five times, then the average frame rate among the five runs was chosen and it was written in the final results. All games were installed on SSDs.

Test results

Special features

Before we go on to discuss our performance results, there is something to note: During our testing of the 20-threaded E5-2660 v3, it turned out that in some projects that use a small number of threads (4-8), it lags significantly behind the obviously weaker E5-2630 v3 chip. We looked into this peculiarity and found out that disabling Hyper Threading (virtual threads / SMT) often has a positive effect on the final performance of such a multi-threaded CPU in games. Thus, we decided to run additional tests E5-2660 v3 processor with Hyper Threading disabled.

On the graphs, this chip variant will be labeled as “Xeon E5-2660 v3@3300MHz 10/10”, and the normal variant will be labeled as “Xeon E5-2660 v3@3300MHz 10/20”.

Assassins Creed Valhalla

Maximum graphics settings, 720p resolution. Built in game benchmark.

In the latest installment of the Assassins Creed franchise so far, titled “Valhalla”, the 10-core, 20-core Haswell shows very little advantage over its 8-core sibling. However, just switch off HT and the 2660 v3 takes the lead, beating the E5-2630 v3 by quite a margin! Actually, this is what we wrote about in the “Special features” section. In “Valhalla” 10 extra threads are detrimental to performance. On “pure” 10 cores some games run much better.

Battlefield 2042

Max graphics settings, 1080p resolution. EXPOSURE map.

Ray Tracing – ON

“Battlefield 2042” with ray tracing activated responded quite well to two additional Xeon E5-2660 v3 cores. Here the gap of the 20-threaded processor was 10% in average FPS and 11% in minimum frame rate. In addition, it should be noted that with RT BF2042 performed slightly better exactly on the processor without Hyper Threading technology.

Ray Tracing – OFF

With “ray-tracing” deactivated, the E5-2660 v3’s lead over the E5-2630 v3 shrinks to 6% in average FPS and 9% in minimum FPS. Unfortunately, these figures are not worth upgrading from an 8-core to a 10-core chip for.

As for the standoff between the two versions of the 2660 v3, here came out parity: both the 20-threaded processor and the 10-core, show identical results, and the difference in rare events is at the level of uncertainty.

Cyberpunk 2077

Maximum graphics settings, 720p resolution. Built in game benchmark.

Ray Tracing – ON

Cyberpunk 2077 with ray tracing shows very contradictory results: on the one hand, E5-2660 v3 showed more confident minimum FPS, on the other hand, 20-threaded chip failed to beat 16-threaded E5-2630 v3 on average frame rate. And that doesn’t sound like a flaw, as we performed several additional benchmark runs where the previously obtained results were repeated time and time again. Nevertheless, objectively, minimum FPS is much more important than average, so, it is more logical to choose E5-2660 v3 for CDPR studio creation.

But with HT disabled, E5-2660 v3 doesn’t look so good: its minimum frame rate value is significantly lower than that of E5-2630 v3.

Actually, according to our research, Cyberpunk 2077 is able to adequately use up to 16 threads, so if the chip has a large number of execution units, it is possible to reduce performance. But here we should make a remark: this only applies to processors based on old architectures like Haswell.

Ray Tracing – OFF

Disabling the notorious ray tracing makes some adjustments: the E5-2660 v3 is now equal to the E5-2630 v3 in average FPS and even further off in minimum FPS. With HT deactivated, the 10-core chip showed an unbeatable average frame rate, but also failed miserably at minimum FPS, which essentially puts an end to disabling auxiliary thread processing technology for Cyberpunk 2077.

Destiny 2

Maximum settings, 720p resolution. “Deep Six” map from “Gambit” mode.

In Destiny 2 multiplayer session shooter up to 8 threads are actively used, that’s why getting advantage by means of additional cores in this game is almost impossible. But Bungie’s creation responds rather well to level 3 cache increase and thanks to this feature, 10-core E5-2660 v3 is still ahead of its 8-core competitor. Nevertheless, it should be noted that the 20-core Haswell did not get a really serious breakaway.

As for the opposition of 10 cores versus 20 threads, the difference can be written off as an error.

Elden Ring

Maximum graphics settings, 720p resolution. Running on a horse through the open world.

The situation in Elden Ring has not changed since our last article about E5-2630 v3: Souls-like action is very poorly optimized, so even on 20-threaded E5-2660 v3 with voluminous 25 MB cache third level small freezes often occur, and the frame rate suffers from extreme instability. However, it should be noted that the best performance is shown by the protagonist of this material.

But that’s not what’s most interesting here. Disabling Hyper Threading significantly reduces the performance of the E5-2660 v3, causing the 10-core to lose to the 16-threaded E5-2630 v3.

Far Cry 6

Maximum graphics settings, 720p resolution. Built in game benchmark.

Ray Tracing — ON

Far Cry 6 is not able to adequately allocate the load on the CPU if the number of its threads is more than 12, because of which increasing the execution units may have a negative impact on the final performance. And this is clearly demonstrated by the graphics above: with HT technology activated, E5-2660 v3 is slightly inferior to E5-2630 v3, but once you turn it off, the “pure” 10-core does not leave a single chance to its little brother and reaches 60 fps in ray tracing mode. What an amazing result!

Ray Tracing — OFF

With deactivated ray tracing the situation has changed quite seriously. The 10-core Xeon E5-2660 v3 without HT increased its advantage over the 16-threaded E5-2630 v3 to 15% and beat its own, full, 20-threaded version by the same amount. It becomes quite obvious – Hyper Threading is extremely detrimental to performance in FarCry 6.

Marvel’s Spider-Man Remastered

Maximum graphics settings, 720p resolution. Flying through the city on the web.

Ray Tracing — ON

Chips without AVX2 are not capable of running a game with RT.

A similar situation occurs in Spider-Man Remastered with activated ray tracing. Here Hyper Threading technology also plays a cruel joke with the E5-2660 v3. Without HT 10-core CPU is capable of 20% more performance. Maybe with new patches the developers will fix this problem, but at the moment the former PS4 exclusive works better without HT.

Ray Tracing — OFF

With RT disabled, essentially nothing has changed: the E5-2660 v3 version without Hyper Threading still leads by a huge margin, while its full 20-threaded version has fallen to the level of E5-2620 v3 by minimum FPS.

Marvel’s Guardians of the Galaxy

Maximum graphics settings, 720p resolution. Built in game benchmark.

Ray Tracing — ON

“Marvel’s Guardians of the Galaxy” with ray-tracing activated is much better optimized for multi-threaded processors: here the E5-2660 v3 has a 4% advantage over the 2630 v3 on average FPS, 11% on minimum and 7% on rare events. Still, here even the 8-core Haswell provides an acceptable frame rate change, which makes the 20-threaded E5-2660 v3 seem redundant.

As for the confrontation between the two versions of the our hero, it turned out to be actual parity, with a slight advantage of the full-fledged E5-2660 v3.

Ray Tracing — OFF

Without ray tracing, the frame rate increases on all processors without exception. However, this does not significantly change the power balance and the gap.

Saints Row [2022]

Maximum graphics settings, 720p resolution. Driving around the city.

Ray Tracing — ON

In “Saints Row” reboot with activated ray tracing, changing from E5-2630 v3 to E5-2660 v3 will not bring you any noticeable change in frame rate, at the same time, if you disable Hyper Threading in the 10-core, you “squeeze out” an extra 7%, which is not bad.

Ray Tracing — OFF

With deactivated RT events repeat themselves: both versions of E5-2660 v3 are still slightly ahead of E5-2630 v3.

Red Dead Redemption 2

Maximum graphics settings, 720p resolution. Built in game benchmark.

Unfortunately, even at 720p resolution the RTX 2060 SUPER graphics card is no longer enough to adequately test the processors in the second part of the legendary Red Dead Redemption franchise at max settings. Even so, it’s clearly evident that the E5-2660 v3 without HT shows slightly less FPS than the 20-threaded version.

Watch Dogs Legion

Maximum graphics settings, 720p resolution. Built in game benchmark.

Ray Tracing — ON

Finally we got to the heaviest game in our test pack – Watch Dogs Legion. The third part of the popular Ubisoft’s series responds quite well to increase number of CPU execution units, but it lasts only up to 8 cores, 16 threads: 20-threaded E5-2660 v3 managed to outperform E5-2630 v3 only by 3%, and disabling HT in this game on the contrary – though not significantly, but it reduces performance of 10-core.

Ray Tracing — OFF

In less heavy mode without ray tracing, the difference between the E5-2660 v3 and the E5-2630 v3 is almost blurred and no longer makes much sense to the end user.

E5-2660 v3 advantage (10/20) over E5-2630 v3 (8/16) in 11 games:

Game:AVG FPSMIN FPSMIN FPS 1%
Assassins Creed Valhalla5.94%6.98%5.97%
Battlefield 2042 RT9.72%15.63%11.54%
Battlefield 20425.94%9.09%8.7%
Cyberpunk 2077 RT-1.41%1.75%3.7%
Cyberpunk 20770%8.7%5.97%
Destiny 25.08%2.97%
Elden Ring1.75%1.96%2.78%
Far Cry 6 RT0%-2%0%
Far Cry 60%-1.89%0%
Marvel’s Spider-Man Remastered RT2.08%0%0%
Marvel’s Spider-Man Remastered1.59%-1.92%-4.44%
Marvel’s Guardians of the Galaxy RT3.7%11.29%7.55%
Marvel’s Guardians of the Galaxy4.4%2.82%3.92%
Saints Row [2022] RT1.41%0%-2%
Saints Row [2022]2.7%3.33%3.77%
Red Dead Redemption 20%0%0%
Watch Dogs Legion RT3.64%2.17%4.55%
Watch Dogs Legion4.17%0%0%
Средний прирост в 11 играх2.82%3.38%3.06%

As a result, the only games that got a noticeable acceleration from four extra threads were Battlefield 2042 and Marvel’s Guardians of the Galaxy. The rest of the projects in our test suite were not able to utilize such a large number of CPU execution threads.

E5-2660 v3 (10/10) advantage over E5-2660 v3 (10/20) in 11 games:

Game:AVG FPSMIN FPSMIN FPS 1%
Assassins Creed Valhalla2.88%5.75%2.9%
Battlefield 2042 RT2.53%-1.35%3.45%
Battlefield 20420%0%-2.67%
Cyberpunk 2077 RT0%-6.9%-10.71%
Cyberpunk 20774.3%-12%-12.68%
Destiny 20%-0.95%
Elden Ring-3.45%-5.77%-13.51%
Far Cry 6 RT7.14%6.12%4.55%
Far Cry 615.38%15.38%12.24%
Marvel’s Spider-Man Remastered RT20.41%23.08%19.44%
Marvel’s Spider-Man Remastered18.75%15.69%13.95%
Marvel’s Guardians of the Galaxy RT0%-2.9%-1.75%
Marvel’s Guardians of the Galaxy2.15%-2.67%3.77%
Saints Row [2022] RT5.56%5.26%4.08%
Saints Row [2022]3.95%3.23%-1.82%
Red Dead Redemption 2-0.99%-1.12%-2.56%
Watch Dogs Legion RT-3.51%-4.26%-4.35%
Watch Dogs Legion-1.33%-1.79%-1.85%
Средний прирост в 11 играх4.1%2.13%0.73%

As expected, the difference in performance between the E5-2660 v3 with Hyper Treading technology activated and the same without HT is highly dependent on the selected project. For example, deactivating multi-threading in Cyberpunk 2077 significantly impacts minimum FPS, but in Marvel’s Spider-Man Remastered and Far Cry 6 disabling HT is a must.

In fact, the results are in front of you, and it’s up to you to decide which mode to use your processor in.

Conclusion

The test results showed that Xeon E5-2660 v3 is quite ambiguous processor. Of course, in 90% of cases it was faster than 8-core E5-2630 v3, however, often its advantage is not so significant, and to get a noticeable “boost” in under-optimized projects, you need to turn off multi-threading technology Hyper Threading, leaving only 10 cores active.

So, it seems quite obvious to us – it makes no sense at all to change 16-threaded 2630 v3 to 2660 v3. At the same time, if you dare to build a system in 2022 from scratch based on LGA 2011 v3 platform, it is definitely better to choose the 20-threaded Haswell.

Leave a Reply

Your email address will not be published. Required fields are marked *