There is an excellent Cache vs Cores video from Hardware Unboxed, so yeah, extra cache of E5-2667 V4 surely matters: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-0mO4op3bL90.htmlsi=94bechl0rEYQSywr
What would be interesting is comparing 4 core xeon with (e.g) 10 core xeon with last 6 cores disabled. Would the other cores' cache be used by the enabled cores? Edit: There's even Haswell CPUs with 45MB of cache...
@@kansax8253the answer yes and no. The L1 and L2 caches are unique per CPU core but the L3 cache is shared, so if a CPU has 45MB of L3 cache and you leave only 6 or 8 cores enabled, entire buffer of 45MB L3 cache will be available for these 6 or 8 cores.
always keeping, up the best work I am safe to say I will still keep my e5 2667 v4 thanks once again for the work. you do and stay safe bless you always.
Bought a Asus X99A motherboard. What is better for gaming and rendering ? An overclocked 1680 v3 or a 2699 v3 with turbo boost unlock ? I also have a 5820K that can overclock to 4.5ghz on 1.35V Vcore. I plan to achieve 1440P 144hz with an RTX 3080.
Еще одно отличное видео! Без тени сомнения, об исследовании этих процессоров будет снято несколько видеороликов за добрые несколько лет. 2697A V4 и подобные им еще могут удивить...
I also suspect it's a temperature issue. The Tcase of the 2689v4 is 52°c. When I monitor the temperature on hwinfo there is a critical temperature point on the motherboard, reaching 80°c. I just purchased a new cooler for TDP up to 220W, but I think I should have bought another low profile cooler, which helps cool the mobo
I was not expecting the 2689 to be that slow, but its cinebench numbers match roughly what you would expect for it specs, i guess the LCC layout of it affects memory enough where its clockspeed isnt enough to make up the difference in L2/L3 cache performance.
Hello whats the best build for E5/x99 for gaming? Fastest cpu and mobo setup for 150$ max! Turbo mod or v4. (Aliexpress) Please answer what is the best one! Looking to build a good pc! LOVE THE CHANNEL!
MR9A PRO + E5-2697 V3 unlock + 4X8GB KLLISRE + any 6 heatpipes cpu cooler. Remember to fine tune memory like 2133MHZ CL12, and also do reBar + fixed BCLK 100mhz modded bios. Edit: For gaming try it with disabled HT and keep only 10 cores enabled.
Thanks for this video and your generous subs! I was strongly considering a locally available 1680v4 to replace my 2667v3. That high price seems to be not worth it's meager performance.
I have the 2689 V4 and I noticed that in the CPU stress test in AIDA 64 it reaches 3.7ghz on all cores. In the FPU test it reaches 3.2GHz on all cores. Another thing I noticed is a high consumption reaching 135w in the FPU tests.
As an IT specialist i can tell that you keep it nice in your analytics and more than once i was thinking the same before you sad it, nice and professional mate and for your situation in your country a long time i was thinking that we didn't need these ways of treating each other that we as people where better than that wat we see now it breaks my hart man. We should do better than that don't you think. We already have so much history to learn from and still.. I wish you the best and hope that every thing will be all right heart felt wishes from the Netherlands
Awesome review, I've got to be honest the results comparing the E5-2667 V4 VS E5-2689 V4 really hurts. I really thought the E5-2689 V4 would of done better. Good Job.
My 2680v4 performs better than my 2697a v4, I suspect my bad mobo is overheating with the latter but the 2680v4 is very capable in my heavy emulation games, and can multitask decently.
@@Miyconst 2650 was my old cpu , i updated to the 2697AV4 as its the best value cpu as i bought it in aliexpress for $40 each (2 units) using them in DELL R620 servers.
@@Miyconst im new to these things thanks for your kind response i seen a deal on aliexpress HUANANZHI X99 QD4 with E5 2650V4 in 58$ thats why im asking.
I have 1650 v4, 1660 v4, and 1680 v4 and had 2667 v4. I can attest that for workstation and desktop usage they seem to be more agile than 2667 v4. I learned that their memory latency is better than 2667 v4 but this is the only difference that I've noticed. The 1660 and 1680 fared better in the userbenchmark and passmark software than 2667. I can only attribute the better performance in games of the 2667 v4 to its larger cache.
Usebenchmark is a complete joke or a test, PassMark is also very biased. The modern games load several CPU cores even if they rely on single core performance, thus the CPU does not hold the max turbo frequency.
Wouldn't Overclocking 1660 v3 be better? even better for competitive games like cs2, it would be cool to see 1660 v3 vs 2667 v4 vs 2697 v3, hugs Brazil my friend
@@Lucas-Vitaloverclocked E5-1660 V3 consumes shit lots of electricity, requires a strong VRM, PSU, and cooling, while the performance is not that great due to the limited amount of cache, only 20 MB.
@@Miyconst Overclocking has little to offer to my usage that is not gaming. But I find it very interesting for CPUs like 2697 v3 if your are a gamer on a tight budget.
Got an Asus ROG Strix x99 gaming and now looking for a CPU, was about to order the 1660v4 and saw the multiplier is locked. So I guess the V4 processors are out of the question if the motherboard has the ability to overclock? So for gaming should I go for the 1660 v3?
Доброго дня! It was already written here in the comments that 2689v4 has something strange with the Tcase=52°C. I've never seen such a low value in any processor. Despite the TDP of 165W. Maybe the processor reaches this temperature and goes into throttling? Can you check it?
Hi friend, nice video as always, i had a question for you, can you tell me wich E5-v3 Processor you think or consider is the TOP 1# for gaming, including overclock and turbo boost on your bios program greetings
Are you planning on doing an review of the Qiyida X99 D4 motherboard? it comes with C612 chipset. The original BIOS is locked for overclocking, but an brazilian youtuber managed to modify the BIOS and unlock the overclocking features.
The E5-1680 V4 is $45 USD, great performance at this price for non-modifiable workstations. Do you disable Spectre and Meltdown? I notice it helps performance.
It's a six core but I bought 2 xeon E5 2643 V4 for 10$ each and with 3.4ghz base clock and 12 threads really sound like an alternative to 8 gen i5, the whole mb + cpu + ram for the price of a cpu or even less is really noticeable, the 1660 V4 would be its 2 extra core counterpart but it's 20$ more, yet still a good price. I only wish there was tpm 2.0 for those x99 motherboards...literally would last another 10 years easy
More MORE MORE! Right now I am running Z440 with a e5-1680 v3 overclocked to 4.29 ghz and a 1080 ti with 36gb server ram even though I have a machine with a 3080 ti. I like getting 160-200fps medium high (epic view distance) settings on Fortnite with a system that cost me less than $250.
One thing that should have been mentioned is that 16 series CPUs have Intel Turbo Max 3.0 technology that the other cpus in the test don't. This is a big plus, but only if you have and older version of Windows 10 (1809 or earlier) and activate it with the proper driver. Maybe you could test these CPUs with turbo 3.0 turned on.
It looks like the E5-2689 v4 was power-throttling on your board. It has a TDP of 165 W, which Chinese boards might not be able to handle correctly. I had one a while ago on an MSI X99 board, and it was able to maintain 3.7 GHz on all cores during AVX workload. It could also be that mine was a QS (QK7Z), but I doubt it because QSs are almost always identical to retail.
I also suspect it's a temperature issue. The Tcase of the 2689v4 is 52°c. When I monitor the temperature on hwinfo there is a critical temperature point on the motherboard, reaching 80°c. I just purchased a new cooler for TDP up to 220W, but I think I should have bought another low profile cooler, which helps cool the mobo.
What an array of results. What was most interesting to me was the way the 2667 and the 2689 seemed to go back and forth between which was faster. It really does come down to the game and use case but I genuinely suspect that binning for this 1st generation of 14 nm silicon was a massive factor. These were the 1st of the 1st and we all know how long Intel lingered on the 14 nm node. So Intel clearly felt that they could get a whole lot more out of 14nm than what they got out of this generation. So it stands to reason that some of the chips are just better than others and possibly significantly so. LTT also recently did a whole series on binning and they found that even current generation chips can have swings as wide as 6% Overall I think that either the 2089 I picked up was decently binned or the 2090 I had was poorly binned. I actually think the later because I could never get my 2090 to turbo much higher than 3.1 ghz(which is far from the rated max of 3.5) and my 2089 turbos to 3.68 most of the time (which is just a hair less than it's rated max of 3.8). If I could have gotten my 2090 to turbo to 3.5 or even 3.3 I think I would have been a far happier camper. =) I too also saw some funky memory timing on the 2089 that seem to hold it back. My 2090 seemed to have slightly better cache latency, for example. So I think your points about the memory behavior and differences in memory controllers are extremely compelling in general and I really wish there was a way for all of the hardware reviewers to band together for some massive group test to help determine what is actually going on in some of these games with some of this hardware. I don't just mean the lga 2011 v3 platform. I'm talking about software/hardware optimization in general. The 2689 never seemed to really pull ahead of the pack and at times it seemed to be oddly slow but overall it did very well and seems like a good compromise of various performance types. My main question from this series of test is what was the typical in-game clock speed for each of the three cpus? My 2089 seems to really like being just under 3.7 ghz in most games with one core near maxed-out and the rest at about 30% An interesting thing in general seems to be that the all core processing power of a cpu seems to affect the minimum frame rates and the the single core speed seems to affect the average frame rate. To me this is somewhat akin to horsepower vs torque on an engine. The min frame rate is usually measuring moments when the game had to load something or compile a shader, so the gpu had to wait a moment to keep rendering frames. These events seem to be tasks that benefit from multiple cores readily. However the main game thread is what determines the frame rate when the game is not having to do this extra work. So it is usually limited to the highest speed that one core can achieve. So a 14 core cpu like the 2090 acts like a high torque truck made for hauling huge loads and the 8 core 1660 is tuned more like it's a V8 muscle car. You're not going to tow a huge load very well in the car but your cruising speed and top speed will be higher. Another round of great tests with more interesting results. Thank you for even more hard work and insights and thanks to your friends for the generous donation! I just have to say that my understanding of the fight for Ukraine is that Ukraine(if given the chance) provides so much food and bounty that both Europe and Russia see Ukraine as critical for their ability to feed themselves and maintain the general welfare of their people. As much as I hate Russian and Putin even more for their terrorism, I understand why they think they need to try to control Ukraine. because they're kind of screwed if they do not do something to compensate for their regional isolation. In other words, Russia may be big but she's boxed in and thus behaves like a cornered feral animal on the international stage. However Europe is probably even more screwed if they ignore Russia's invasion of Ukraine because they have more people and are closer to Ukraine and would benefit so much more from Ukraine's economic and agricultural output than Russia ever could. So ignoring Ukraine's suffering is like ignoring the suffering of a neighbor who has helped you many times in the past and will help you countless times in the future. So for the leaders of Europe right now, defending Ukraine is not just a moral imperative. For Europe, defending Ukraine is about defending the future stability and prosperity of all of Europe. (The fact that Ukrainian soldiers seem to have a knack of causing massive collateral damage to Russian military machine is just icing on the cake.) Even in America only half of the GOP's voters have their heads up their ass about this issue and that's mostly because of Tucker Carlson and Donald Trump. The other half is very much in support of Ukraine and most of the rest of the country is, too. I think the GOP will come back to the table and I think American aid to Ukraine will resume. At least if we have any sense it will but I'm almost certain that cooler heads will prevail. Thanks again and Glory to Ukraine! =D
It is posible to OC E5 1680 V4 to 4000+mhz on X99-MR9A? and it make sense buy V4? or better V3 with Turbo Unlock, which option will be better to X99-MR9A for gaming etc?
@@Miyconst There you overclock E5 1660 V4 ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-RIVKU85AC6Q.html and there ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-maWMtmCYOTQ.html
Ive been wanting to a 2689v4 for a while but seeing that performance numbers are almost identical as the 2687w is somewhat of a disappointment especially since the price remains as high as it is.
Wow - very disappointed in the E5-2689 v4. I'm pretty surprised it wasn't maintaining its all-core boost clocks - it should be 100-200MHz higher than the E5-2667 v4 according to the Wikipedia page. I have a hard time believing the two extra cores on the ring bus leads to enough contention to cause problems, but it's possible! I can't think of anything else that would cause that... other than the 2689 v4 being a MCC die with a lot of cores disabled instead of a LCC die with all cores enabled. But I'm pretty sure based on the spec sheets I've seen it's a LCC die. Very interesting results though - appreciate the video as always.
Loved the video, but could you test Fortnite with some of these cpus? I'm mainly looking for the results with just the 1660 v4 vs modern ryzen cpus and modern intel, Only because playing fortnite at 1080p with RTX 3070 and Xeon E3 1271 V3 I am only getting around 40 fps with ray tracing on and DLSS at performance, and overall the game just feels CPU bottlenecked, I feel like the graphics card can hold at least 60 as it hovers around 40-50% utilization and im thinking that's due to the CPU as sometimes I have texture loading issues, but I love your videos. Would really appreciate if you could test this as you have an equivalent GPU the RTX 2080 is a good match.
Hi there. Xeon E3-1271 V3 is unfortunately very weak for gaming these days. I have a Fortnite gameplay with E5-2697 V3 it can help you understand the E5-1660 V4 level when you know the overall difference between these CPUs: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-cWZXmOczIpc.htmlsi=53rBrbiVQ3PBt4hc
I've got a 2699v4 paired with 3060 12gb and in fallout 76 I get less than 40fps on high and between mid 20s and mid 30s depending where I'm at. It's in HP z440. I'm pretty sure the HP z440 is bottlenecking my CPU... maybe
All AliExpress sellers are questionable, all of them treat consumers as garbage at some point, I can't confidently recommend either one of them. Still, welcome to join my discord if you want some specific links where I would buy.
отримав, протестував, хороший проц за 24 бакса, нажаль китайці вже підняли ціну до 40. практично як 2667в4 та має бонус - працює андервольтінг у intel XTU на моїй MR9Pro, при -90мв споживання системи з розетки у аіда стресс тесті знизилось з 155w до 138w. З ресайзаюл бар та рх6600 дуже збалансовий конфіг для 1080.
2689v4 deveria ter performance melhor, porém o que vemos são núcleos operando em baixas frequência e tornando seu desempenho aquém do esperado.0 2667v4 continua uma excelente opção devido o cache L3 e frequência estável nos 3.5ghz. 1660v3 overclock chegando fácil nos seu 4.5ghz seria interessante para jogos apesar de arquitetura mais antiga. Excelente vídeo...deve ter tido um trabalho grande! Abraço aqui do Brasil. Força para o povo ucraniano contra a tirania de Putin!
Thank you for the support! E5-1660 V3 consumes a lot of power when overclocked, it requires expensive motherboard, expensive cooling, expensive power supply.
Interesting review. I'm of the persuasion that a mainstream motherboard is the way to go. My Asus X99-II with an E5-2680v4 with 96GB ECC ram, gets a consistent R23 score of 12200+ A good motherboard also provides better utilization of sleep function, remote management and other features. Although the cost of a used X99 mainstream motherboard may be hard to justify just now. My experience anyways. Cheers.
2697Av4 and 2690v4 are doing well and are the best compromise for me .. right next to the 2667v4 . The 2689 ist a big and expensive disappointment indeed
Love the videos❤ was going to make my new pc with the 2697v3 before i decided to make it on AM4. Keep up the good work and of course: long live ukraine
Thank you Kostiantyn, another interesting video. I'm glad that you still have time for some X99 science. Despite these disappointing results, I'm always surprised how close they come to Zen 3 given the four year age gap. But it looks like 2667v4 is still an excellent budget option with a well priced motherboard. Cheers. ~386DX.
@@gabrielecarbone8235 yeah, that would work as well.. but back then Intel still felt like a king and tried to provide absolute minimum with absolute maximum profit.
Te felicito por tu trabajo. Eres mi referente en este tema de procesadores y placas chinas para Xeon ! He comprado diferentes motherboards y procesadores en aliexpress y debo decir que estoy muy conforme con sus rendimientos. Nuestro apoyo a Ucrania desde Argentina
@@oskrtellez3990 un combo más mucho más económico y muy buen rendimiento es placa Huananzhi BD3 , xeon 2667v3 y memoria Ddr3 2133mhz. Muy buen rendimiento por menos dinero
i hope someone send you a 2679v4 for testing. Its WAY too expensive (and also rare) to even justify buying it but id love to see how it fares against everything else since ive never seen proper benchmarking of it.
Man... I wish Aliexpress not banned in my country T T My local e commerce put insane price for V4 cpu and i want 2667 v4 and 2680 v4 Only worth one is v2 but ddr3 lol Welp ended build 2667 v2 32gb ddr3 1833 for 80$ At least the single core performance ain't that crap
Can we let v3 and v4 Xeon rest now? 5700x3d and 7500f are the way to go for saving money and electricity. Good 1% low fps will require RAM OC or a lot of cache.
Excelent video. I have download your Mi899 file, however, I could not find my motherboard, Qiyida H81, chipset H81. Is this model not included in the Mi899 file?
Qiyida is an AliExpress store name, there are lots of different motherboards under this name, I need to see a picture. You can send one via my Discord.
Wouldn't Overclocking 1660 v3 be better? even better for competitive games like cs2, it would be cool to see 1660 v3 vs 2667 v4 vs 2697 v3, hugs Brazil my friend
E5-1660 V3 consumes too much power when overclocked. It requires expensive motherboard, expensive cooling, and expensive power supply, while the final result isn't that great due to the small 20MB L3 cache capacity.
Yes it would but not with any motherboard and not every stick. Some motherboards support only ECC REG memory and 4/8-bank regular non-ECC modules are not supported. Please note that 4 banks is not the same as 4 GB.
hello Miyconst, im thinking of making a cheap Xeon build since my main pc died. im thinking of getting a Xeon E5 2697AV4 for around 45-50 euro, what motherboard brand would you recommend from aliexpress?
Hello Miyconst! A customer of mine bought the Xeon E5-2680 V4 and has since then claimed he can't play Valorant on Windows 11. Is there a solution for this? His motherboard is the Machinist X99 MR9A PRO MAX. Thanks in advance!
hey myconst i its a bit confuse with xeon power usage in your video i see xeon use 16 watt to 60 watt only ? but in net i see 100 watt ++ 🤔, is it because undervolt? and in you video in power consumption section xeon use around 400 watt system wide , is it mostly your gpu(7900xt) ? so gpu use 230- 300 watt, rest of the system only use 100 watt? so i have x5670 in motherboard p6t deluxe (16 year still run strong 😁) i overclock to 3,4 ghz 1,14xx volt overclock system idle watt consumption(system wide) : 130 watt gaming consumption : 200-215 watt , GPU only use 30 - 50 watt so without gpu only use 180 watt system wide stock system idle : around 110 watt so my overclock take 15 - 20 watt so if i buy for example 2680 v4 will i more save energy than my system now ? sorry too long thanks hope you answer my question 🙏
The power consumption reported by software is not accurate. Chinese X99 motherboards mostly have faulty sensors. I report power consumption measured with an external wattmeter.
I have Gigabyte Ultradurable X99 just replaced cpu at E5-1660 V4. I tried stress cpu with cpuz but highest clock speed is 3,2Ghz. Spec’s promises 3,8Ghz is there something wrong
Ok i have always believe that when you stress or bench cpu with CPUZ sofware cpu you can see cpu max frequency. I want to try bios mod but my english language & knowledge is not enough to do so.
Do you know the reason for why the E5-2697 v3 is often more performant than higher frequency v4 chips? It's not like a game will use all those 14 cores on the 2697v3.
That's because of the Turbo Boost Unlock on E5-2697 V3. Intel's implementation of Turbo Boost frequency lowers clock if several CPU cores are used, and it doesn't matter if the cores are slightly used or 100% loaded, that's why both V3 and V4 drop clocks in games but TBU forces V3 to maintain the higher frequency at all times.
Hello, the 2689V4 at 5:04 in the video is running at 3700MHz, and at 5:11, it quickly drops to 3200MHz. The CPU usage and its consumption are low, considering the TDP limit. What is the reason for this frequency drop?
It's the Intel's logic of Boost Frequency. The more cores used the lower the frequency and it doesn't matter if the cores are fully or partially loaded.
The 2689V4 all core Non-AVX Turbo Boost is 3.7ghz, and although I don't have the 2689V4 figures (and they're not on the Intel charts since 2689V4 seems to be some off-map thing), I'd guess the all core AVX Turbo Boost is 3.2ghz across the board on the 2689V4..
Дякую. Десь 5 мiсяцiв тому придбав собi 1660v4, тодi не було нормальних оглядiв його потужностi, та зараз я бачу, що за 24 доллара це взагалi нормальний процессор. Однак моя материнська плата Huananzhi x99tf iнколи ловить синiй экран в тривалих тестах ОССТ, що на старому Е5 2678v3, що на "новому" 1660v4. У декiлькох iграх я вже помiтив рiзницю: статорiв у fallout4, FO тепер нема.
@@Miyconst Чесно кажучи, хочу перейти на плату як у вас, Machinist Mr9A pro, у мого хуанана на старому процессорi якось перестав працювати 4-канальний режим ОЗУ, на новому ще не перевiрив. Ось востаннє перевiрю ОЗУ, запозичивши у друга кiлька планок, та перейду, мабуть. Як ви оцiнюєте моє рiшення?
@@ИванЮрченко-ф2збільше нагадує лоторею, якщо Huanan ще працює та виконує свої функції - не треба його міняти, хай працює доки не помре. Моя X99-F8 де помер один канал ОЗП досі працює в ПК, що я склав для родичів.
Prayers for Ukraine and best of luck to your channel. I'm hoping to see your channel grow and expand into the RU-vid tech genre. Your style of presentation is clean and direct which is a blessing. Thank you.
