Видео

Dragon's Dogma 2 тестил на очень ранней версии, сейчас должно быть лучше. Но тогда уже было 65 фпс средних где-то. ruclips.net/video/JquH1tLq9YE/видео.html

В стоке можешь посмотреть 10700F на MxBenchmark PC или что-нибудь порыть на RUclips по 9900K/10700K, или даже 5700X, но спойлер: намного медленнее. У нас ещё скоро будет сравнение в догме на NeoChannel с 5700X3D

Have you tried tightening secondaries/tertiaries at 3200MHz? It should be capable of at least that shouldn't it?

Thanks for your opinion bro :) ACIII was my entry to gaming on my old GT540M laptop, then ACU inspired me to build the GTX 970 rig. I waited for these games like a fanatic, a very important part of my childhood. So I devoted that video to my golden era so to speak, and tried to showcase the prettiness of each game. People always seemed to hate smth about the series, but I feel like Ubisoft, and AC in particular back in the day, pushed the graphics and open worlds more than most games, if not all. ACU is still crazy looking... Even though I consider the series to be too lasting, weary and kinda grindy, AC Shadows, for now at least, has impressed me more than any open world game since like 2020. Idk if it's just me, but these graphics, dynamic weather, the entourage, all just feels like such a divinely crafted world.

@@HeartOfAdel Yeah there is something to even newer assassins creed games still, i usually do almost 100% completion and they certainly make a decent open worlds, at least in my opinion, sountracks are also super good. I feel that valhalla was the weakest in this new ac era but still pretty decent overall.

@h1tzzYT lately started ignoring all that 100% stuff, because most of it isn't really worth it. What made you think Valhalla was the weakest though? I love that one a lot for the superb medieval vibe, and it is a lot more polished and creative content wise in comparison to Odyssey :)

@@HeartOfAdel Valhalla had decent story arcs but i always missed the story about the Ivor itself, the main character served only as a tool to describe and narrate other character stories and their developments, but not about Ivor themself. I liked valhalla dont get me wrong i just think it was the weakest out of origins and odyssey. The 100% completionist stuff for me i dont mind it, i certainly wouldnt want every game to be like that, but once in a while playing various games its kinda nice, to get a complete freedom and just travel and explore ubisoft maps.

Obviously lol :) It's ram speed and subtimings mostly. But ring and core are also adding their 5-10%.

I've never encountered timings with such names, it doesn't exist on my board. And yep, PPD 0 :)

@@HeartOfAdel i got a question about trefi, is it good to just put it at max 65535? currently at 32768

​@H0tCh0c0-gr3qe if temp isn't a problem - sure. I usually don't push it that high because it makes zero sense at all after 40000-50000, besides maybe a prettier number.

I have cs2 in my previous video. Haven't tested valorant yet. I think I'll have a look at some point

Thanks brother 🙏 What do you mean by settings?

@@HeartOfAdel on the Cpu (vcore, llc, etc etc) memory xmp or manual timings ?

@@HeartOfAdel Sorry bro, I skipped the beginning of the video where you set up the overclock...hahaha xD Congratulations again

​@GazzaniSilva haha yeah, it's in the beginning. Talking about additional settings, it was 5/4.8 ~1.25v load for the video (now 5.1/4.8 1.23v) with LLC5 (set with an offset, AC/DC Loadlines are at 0.18) All subtimings are manually tuned with tweaked training algorithms, slopes. etc

Yes, there should also be 5-10% difference between 2666/3200. Moreover I went for the conservative 5GHz, with a little more juice could gain a couple percent more. So yeah, the gains are crazy. Good tune btw, that's a very fast chip in your hands!

Разгон ядер и кольца с 4.7/4.3 до 5/4.8 даёт около 10%. Значительная часть повышенного потребления результат быстрой памяти.

WTRs 2 потому что WRRD_dg на 24. Можно было сделать 1 при 23, но не к чему. CKE стоит на 8 на постоянку, PPD всегда ставлю на 0 вручную. Чисто для красоты видео выставил cke. WR не правильно отображается конфигуратором, в биосе стоит на 12.

Trefi не люблю максимум дабы избежать проблем при высоких темпах (хотя я их не замечал). Стоит 50000 сейчас. Разницы никакой на 65000. 1.55V абсолютно нормально на b-die, сверху дуют два 160мм вентика.

Все тайминги + RTL/IOL + кольцевая шина с ядрами + PPD + оптимизация приложений Windows. Вот откуда такая задержка)

@@HeartOfAdel так, планирую продать свой i7-8700K и перейти на i7-10700K + msi tomahawk. про эту МСИху пишут, что хорошо разгоняет как одноранги (в том числе 4 модуля), так и двухранги. пока что у меня 2x8 gskill flareX, память способна в 4533 мгц, но из-за процессора уперся в 3600 мгц 14-14-14. да, на однорангах, такое бывает :) да и 6 ядер мало уже, в том же TLOU если у тебя меньше 8 ядер, тебе там делать нечего. да, я энтузиаст и беру железо именно из-за самого процесса разгона. "апгрейд" обойдется в 2500₽, да. а беру комет из-за того, что там контроллер памяти лучше, материнские платы лучше и более просты в разгоне. хорошие питальники, даже в томагавке он неплохой. а на Z390 именно на нем экономили, а про 370 я вообще молчу. там у всех вендоров поголовно 4 фазы с даблерами. на 1151 что-то ниже максимуса/экстрима от асрок брать вообще нет смысла, если не хотите упереться в материнку.

@@sysagent так! xD Удачи при переходе, апгрейд будет, и не в скобочках. От вашего 8700К на 10700К похожий на мой 9900KS будет 15-20% прироста во многих играх, только задержки будут хуже потому что это обрезанный 10-ти ядерник и маловероятно что будет 5.1/4.8 на 1.23V в нагрузке. Зато холоднее будет.

Здарова, думаю должен в целом, если 2х8 память, но никак не с 2х16. Преступление не разогнать такое железо) Но напоминаю что дело больше не в частоте памяти, а в ужатости субтаймингов, и конечно же скорости кольца. 5.2ггц будут невероятно горячими с быстрой памятью, возможно придётся отпускать.

@@HeartOfAdel я думаю 5.1 оставлю и колько 4800 зафиксирую,у меня ревизия проца 20 года 42 неделя,удачный в целом,а память да я понимаю что дело конеч но в таймингах ,на дня буду гнать и ужимать вторички третички , главное разобратся в этом всем грамотно и не сжечь все к хуям кошачьим аххахахаха

@@HeartOfAdel у меня 4x8gb плашки

@@GbcSquad да, хороший экземпляр) разберёшься, все там понятно. Как это 4х8? Z370 Apex же двухслотовая...

@@HeartOfAdel да это понятно,у меня просто есть 4 плашки, буду из них выбирать какие погонятся лучше,(у меня раньше z370 e gaming была, на ней они даже 4000 не взяли,буду на apex теперь пересобирать

Вроде бы asrock z390 это t-topology? Если да, то будет лучше 4 плашки по 8, но 3900cl19 уже норм. Чтобы достигнуть 4266cl16 с жёсткими таймингами мне нужно было очень постараться. Что имеется ввиду под "нужно арешек"?

@@HeartOfAdel ну апекс материнка, Арех) да платформа уже старая и смысла нету инвестировать в ещё один кит хороших бидаев по цене ддр5 ради 5%. У меня 9900к 2019 года в стоке работает на 1.184мв 4.7 ядра, 4.5 шина и вот память 3900цл15. Думаю на че его менять, по видео он так-то не особо слабее 12600к

@@Pixrl3491 apex плохо гонит дуал ранк, 4 слотовые asus z390 это верхушка разгона 32гб. Я недавно тестил z390 gene и она категорически отказывается запускать выше 4000 с 2х16, apex точно такой же. А моя плата может сидеть на рабочем столе даже на 4533cl16 :) У тебя очень хороший 9900К и не вижу смысла его менять года 3, процессор хороший. Вообще минимум какой нибудь 13700К, но и тут только если карта прям самая новая

Thanks, that's very good to hear! Exactly the type of comments this video was made for ;) You should add 2 sticks, 32gb are becoming the standard anyway, then max that baby out. If you have max tight subtimings it'll be around a 5-15% improvement (+from dual rank), so it's worth it overall. Max OC 9700K is still very good though.

I'm extremely curious myself, will check out later when it's pirated lol. I think it'll be close to stock 5800x3d/12700k. Or maybe even you could help me test it through your account. I've seen 5800x3d get 40-50 in the center of that one big city, are you sure 12700K, even with a perfect tune, can get to 80-100fps with RT? Maybe it's in the woods?

@@HeartOfAdel I bought an offline account on eBay. I didn’t buy the game. I can email the guy for 2 authentication if your curious. Here’s the info DragonsDogma2PC 2BBk@5LQu6B as for my FPS, it only drops below 70 when theres alot of NPCs nearby in daylight. The frametimes are pretty bad though, I am going to see if tweaking reframework can improve it

@@jameyt1 add me on discord please discord.com/invite/g7pcDSg9

Yeah, 9900K OC is really good and smooth. And nope bro, it's not too late. If you find 2x16 3200cl14 kits at a really good price, you'll have a generation faster cpu with top-tier desktop snappiness

Yep with 2080ti the gain is marginal, but once you move to 3080 Ti/4070S territory it becomes more apparent. But bare in mind I intentionally forced cpu limits here. In normal scenarios out of the box i9-9900K is already mostly enough for 3080/4070.

@@HeartOfAdelI turned on ReBAR on 2080ti by customizing the BIOS ROM. I will try the ram OC

@@pelukeiro959 oh, you're here as well :) Have you tested it? Is there a performance difference in supported games? Absolutely try ram oc if you want to saturate this cpu.

I appreciate you watching my video, thanks ;) There's almost no data on what a complete oc does. To the point that even enthusiasts don't know the importance of memory subtimings.

Your setup is very good, especially the value of that 9900KS if you actually managed to get better silicon. I'm not sure if that board is daisy chain or t-top, can't say if 2x16 or 4x8 is more optimal. In any case your priority should be the tightest subtimings (3600-3800mhz would be a good entry) and preferably sub 35ns latency. When you add a ring oc it all compounds leading to such gigantic gains. Bare in mind though after a full ram tune managing core clocks and wattage will be more difficult 🙂

Honestly I myself was the kind of guy that only tuned primaries. Until that very moment when I watched buildzoid's video on tuning rrds/tfaw, and saw a huge gpu usage increase in Watch Dogs Legion even at the same frequency. Mind blowing that was. You can join our little discord chat related to b-die oc on coffee/comet lake discord.com/invite/7Qsq495q

@@HeartOfAdel What do you think would gain the most performance (in games), 3800cl14 or 4266MHzcl16? Latency for the 3800 is 7.36ns and 7.50 for the 4266. Some people claim this true latency is decisive, other claim higher mhz helps. Some say it depends on the application.

@@ISFRedSandman provided both run tuned subs, clearly that'd be 4266cl16. The bandwidth is just higher, their true latency is almost identical. But frequency alone doesn't matter as much as tuning all the timings, the gain would be like 3-5% at best.

Нравится мне энтузиазм нашего человека, на западе реже находишь такое желание что-то затюнить/выжать, от того и самые просматриваемые каналы не только забивают на это, но ещё и ставят павер лимиты = бесполезные цифры. 9900К с одной и той же памятью уже немного лучше 5600х, с полным разгоном у него лучше редкие события чем даже у затюненного 5800x, с похожим средним фпс. Это всё из-за монолитности и низких задержек памяти, которые ещё комп делают в целом немного отзывчивее. ruclips.net/video/rbuLT5cYdPQ/видео.html (тут это хорошо заметно, 1 и 0.1 чаще ближе к среднему фпс, и плюс здесь ещё тайминги не полностью ужатые у 10700К/10900К, можно лучше) + Я также прикрепил ссылки на разные сравнения в первом комменте, повторил 4 игры из тестов Neo Channel чисто чтобы сопоставить с 5800x на тех же 3200 cl16, но 4 ранга. Спасибо что поделился идеями, в целом думаю можно это провернуть. Не знаю справится ли плата с 4000мгц, с двумя рангами точно нет. Насколько проц будет "раскрыт" сложно сказать. Даже топ Z390 платы не прямо всё могут выжать из него (к примеру мой кп может 4400-4500мгц). Плюс, большая часть прироста идёт от комбинации разгона кольца + памяти, без кольца будет боттлнек в коммуникации между ядрами. Если ты поставишь 4.6-4.7ггц соответственно кольцо уже не поедет выше 4.3. Но в целом проц будет нормально раскрыт думаю. ruclips.net/video/zi_77tdRw0A/видео.html (27:50)

@@HeartOfAdel мне не привыкать. еще со времен 1150-го вышло так, что дешманская патриотовская память (не зелень, если что), какие там чипы вообще понятия не имею - нигде информации не увидел, спокойно разогналась со стоковых 1600 до 2334 мгц (использовал strap 125 mhz). первычные тайминги неплохие - 11-13-12-20 CR1. остальные тайминги там так себе ужимались, но поджал до упора в свое время - дополнительные 8-15 fps. но сам факт, что на полнятие частоты память очень хорошо реагирует и работает стабильно, при том, что вольтаж стоял вообще 1.6 вольт, насколько вообще помню! тем временем всякие фурри начинали буксовать уже на 1866-2000. попадался один раз комплект beast (красный), 4+4. ранговость не помню, это было два года назад. и даже на 2000 мгц я ловил синьку в тесте аиды (даже не стресс, а бенчмарк "кэш и память"). и напомню, что патриоты были вставлены по схеме 8+4+8+4 (24 гб), dual rank. и это даже несмотря на то, что первые 4+4 были куплены вообще в 2015-м году, а дополнительные 8+8 гб аж в 2022 :)) то есть, там вообще совершенно другая партия. и такой нехилый разгон для полностью забитых четырех слотов. даже запускал ее на 2666 мгц (ddr3), прям какой-то лютой нестабильности не было, тест кэша и памяти прошла, но скорость была значительно ниже чем на 2400 мгц, примерно как у 1866 мгц, лишь латентность ниже. понятия не имею почему так, может особенность хасвелла, а может тайминги нужно было поправлять, а я тогда в такие дебри еще не лез. да даже если 4000 МГц не возьмет - пуска будут 3866, тоже вполне нормально, особенно по сравнению с мутантом и его 2800. вообще ни о чем для такой памяти. нафига я платил 7.000 за неплохую разгонную память?

​​@@HeartOfAdelтак, у меня появился вопрос. b-die синглы 32 гб (8+8+8+8) разгонять легче или также сложно, как и 32 (8+8), но дуалы? сейчас я перешел на b-die память, подвернулась хорошая возможность, правда лишь 16 гб (8+8), но пока хватает. поэтому такой вопрос. а то насколько я знаю, b-die дуалы гонятся до 3600, ну до 3900 максимум. i7-8700k 5000/4800, G.skill Flare X 4400 мгц 17-17-17-36 CR2 1.488v + обдув. материнка досталась по дешману MSI Z370krait, в плане питальника она так себе, но выше 5 ггц гнать не планирую, но их она вывозит. перешел на Z370, так как заметил рандомные ребуты у asus 270-A, притом без нагрузки. все откинул, остались лишь материнская плата, и мутант, но именно материнка под вопросом. впрочем, она сейчас уже выведена, и это другая история.

​​​​​@@K961KKKот ранговсти разгонный потенциал самих плашек особо не меняется, зависит от отборности. Обычно те которые заканчиваются на A (GTZRA к примеру) у G.Skill будут разгонятся лучше всего вплоть до 4400cl16+ вне зависимости от ранговости. Разгон зависит от топологии материнки. Daisy Chain гонит две плашки лучше, T-Topolgy 4. MSI как я знаю в Z370/Z390 использовали daisy chain, значит 2x16 лучше чем 4x8. Высоких частот я бы не ждал, но максимально низкие вторички/третички/Rtl,Iol можно будет достичь на 3200-3600мгц + разгон кольца (самое главное). Лучшие платы для разгона двуранга памяти в то время это t-top платы asus и может gigabyte, ну и конечно же 2 слотовые.

@@HeartOfAdel да имелось в виду то, какой режим будет легче для контроллера памяти. 2x16 или 4x8? по факту одно и тоже?

Спасибо за просмотр, приятно слышать :)

Thank you for watching! You should see a big improvement after tightening subtimings in many games. But sr will probably not clock above 3600 on your board, 4 sticks should. And sr 3600 is also far from the limit of the IMC which can take it to 4600mhz+. So to maximize your 8700K you'd want 4 sticks of these Asgards. I assume with 5ghz core/high ring and fully tight dr 4000mhz+ b-die that chip is getting close to R5 5600X3D, probably a lot snappier overall. 8700K is ahead of its time, you can rock it for 3-4 years more like that. Ring does scale better with fast ram, you're right.

If your interested in tweaking W10 you can get the process count down to approximately ~40 without messing up the OS/the event viewer. If your not playing games that need specter+meltdown turn it off in Windows, it benefits older intel CPUs. You might need to change a file in system32 as well to make sure it turns off (mc update_intel I believe) Easiest ways to reduce process count would be by clean installing GPU drivers with NVCleaninstall, disabling services you don’t use/need, and removing the windows search with startisback (much better than windows search).

@@jameyt1 have you noticed any gaming/snappiness gains after all that..? I tried that one app that disables meltdown+spectrum but didn't notice any improvements.

@@HeartOfAdel check CPU-Z see if the single core score changes. Usually disabling those mitigations helps a lot of older intel chips. Tech yes city did a benchmark long ago with the 10900k, it helped intel platforms but not AMD

@@jameyt1 I think I did that as well. Literally no changes in scores. I wonder if 2021 produced i9-9900Ks had something patched up in them.

Damn I was waiting for that comment. What do I do now I'm caught? 😂

Literally not true, gained 45fps on my 9700K from 3200c16 to 4000 flat 15 and subtimings. I made my own benchmark run that I ran 3 times and took an average. Idk where you're getting "fake" from.

@@HeartOfAdel You deleted my comment, why? Do you know that I'm right, right?

​@@tvcultural68 I never saw your comment. I ain't Hub to escape reality.

@@tvcultural68 there's a possibility youtube saw it as spam and deleted it automatically

Depends on the card and settings/dlss. With 4090 absolutely would. With 4080(S), probably not.

Bare in mind that settings in these video aren't real world, dlss is pushed really aggressively to cause a cpu bottleneck. 3080 and 9900K are perfectly balanced at 1440p + DLSSQ. At 4K with 3080 that cpu is overkill.

I originally had FC5 included as well to compare with 5800X from Neo Channel. Removed it because it was getting too long. Stock is 177 avg, 136 min, 1% around 130. Tuned is 217 avg, 170 min and around 160 1%. Around 23-28% overall, not the most sensitive game, but better than FC6.

I've found that FC games are more bandwidth/cache sensitive. Although if you look at the pinned comment and compare to FC6 numbers from sirvikman and i2hard, it doesn't help the 32mb of l3 on 5700X beat 9900K even with 3200mhz ram. (So there is some latency sensitivity..?) They're roughly on par at stock and r7 loses by 5-10% in avgs/lows when both max tuned. 5800X3D is a different story though, it outperforms it by about 10-15%, 5700X3D probably by 5-10% only. Very cache hungry indeed. Nothing well beat SOTTR in that regard though 😂

Yes it's really really good. Lows see the biggest increase, very easy to see it at 30%+. Take about 2-3% off my numbers at 4.7GHz and this is how much ram oc would bring alone. I don't like to separate it from ring oc because these are intertwined and compound on each other.

Would it be possible to achieve somewhere near this with a LP 3600 C18 kit from Corsair ? @@HeartOfAdel

@@TROD.. Unlikely, but it should still reach decent subtimings. Coupled with tuned ring/core, I think you can gain 10-20% from stock/xmp i9-9900K

Yeah, because at the time it was just this ahead of everything else.

14 nm +++++++++ needed to be replaced

@@arel77 replaced with what..?

@@arel77 I just realized you're the dude from reddit ;)

@@HeartOfAdel hehe 🤫 🤫 don't tell no one

What do you mean? This is a cpu test so naturally setting were pushed to avoid 99% gpu usage. In normal scenarios 9900K(S) pushes 3080 very well even in quite cpu heavy games.

And thanks for watching 😊

you clearly don't understand the test, what ubderperform are you talking about? your statement makes no sense at all.

Haha, you... xD

Yes it's correct. Looks like newer games are better at utilizing some parts of the tuned processor. But still there are older games that have a 20%+ gain and there are some more modern ones that don't get above 15-20% improvement. Depends on the engine. I had more games tested than 18 actully, but the video would get too long. Around 25% improvement in AC Origins and Far Cry 5, 20% improvement in Ratchet and Clank and Robocop. Fortnite has 15% better avgs but lows shot up by about 40%.

Newer games have larger amounts of NPCs, geometry, physics simulations, and so on. This increases the amount of data the CPU accesses per frame, which in turn causes more cache misses and a larger pressure on the memory subsystem.

@@noreng9333 that is a good explanation indeed. Sounds about right :)

And yeah, it's quite expected 9900K can only reach it with 4 dimms, rather weird comparing to later gens. But that's just the memory layout of most high end Z390 boards. Even two dimm apex seems to have less luck with dual rank than t-tops.

First off all only run 2 sticks, put the other 2 away. if you truly need 32gb, 16x2 b-die is rather expensive. Luckily you can just buy a 32gb kit and a super cheap kit, remove the sticks from the heatsink and return the 32gb b-die with with the cheap sticks on and you keep the 32gb b-die kit.

I wonder how much my trident z royals sell for (2x16)… they are pretty much god bin b-die

@@jameyt1 Pennies probably🤣 The problem with selling stuff like that is that there's basically no demand for it. The noobs will see you asking idk $100 for 2x16 but then they see on amazon a 2x16 kit for $45 and they gonna thing you've gone insane. Then you have the people that would actually pay for that kind of stuff, which represent a very tiny percent of the population... Additionally, every body who can pay 100$+ for ram is already on DDR5, to put simple, your customers are already on next gen basically, so nobody wants that sheet. Like I mean I would buy that but then again I'm already on DDR5 on my main rig so like 🤷‍♂... Also I never really understood the whole "binned ram" debate. Ram makes such little heat that what difference does it make running it at 1.6 vs 1.7v? I mean I personally have daily-ed A-die at 1.9v since launch and it just never passes 35c like idk man, obviously if you're passive cooled then you're gonna run into issues really fast... Also if this kit eventually dies at this voltage I could care less, its as easy as buying another one, swapping pcb's over and sending it back as defective. 💡. But from the looks of it a-die is not too voltage sensitive, temps yes.

yeah I would imagine, everyone seems to be switching to DDR5 anyway since A-die is getting so cheap now. I spent $400.00 like a dumba$$ on DDR5 A-die when the first kits hit the market, now they are worth nothing. I might have to try abusing the return system 🤣. I would kinda feel bad though @@user-wm3fc1sk1p

​​@@user-wm3fc1sk1pthis is not the right advice unfortunately. Asus Z390 boards are t-topology and reach the best signal integrity with 4 sticks. In that way it is a rather weird platform because the only mobo where 2x16 can go to 4200mhz is the Apex and Gene. And weirdly there's more reports of high ram overlocks on higher end t-topology z390 boards than these two dimm ones.

You're fine man, not losing too much. The difference from 3600 to 4200 isn't that big, around 5-7% maybe. It's all about subtimings. If they're tightened out (which isn't that hard to do at 3600mhz) it's pretty damn fast.

@@HeartOfAdel yeah I tightened the subtimings the best I could. Definitely won't be upgrading until 15th or 16th gen

​@@itsbige8312what card are you running?

@@HeartOfAdel its a rtx 2070 super msi gaming x trio, I know I could be running a better card, but for esports/fps titles I don't really need something better tbh.

@@itsbige8312 then you don't need anything above 9900K. That's already quite overkill for that card in many games, and plenty for competitive fps :)

Hey, I have a hard time telling because I've never seen 5600x in action. I think the difference will be there if you are sensitive like some people have noticed. And if you do a lot of fast microtasks (that's when supposedly there should be more difference) And yes, it's very very snappy. I think the major responsiveness improvement comes from ram/ring tuning, it makes an insane difference in system latency/smoothness, like gaining extra 60hz on my 240hz monitor. I'm currently back to windows 10, and on windows 11 the difference between tuned and untuned 9900K was more noticeable. So I think on win11 latency issues should be more apparent.

I'm also not sure if it's worth it in general, especially if you spend much money. In gaming they're about the same, productivity too. The only upside is that superior latency which not everyone can notice. It's also kinda necessary to have fast ram and good subtimings to unleash the snap and full gaming speed. Have a look at my new video to see the difference.

@@HeartOfAdel My 5600 isn't less snappy but I do notice some delays while opening file explorer from taskbar, device manager etc considering I've even optimized debloated and tweaked my windows for latency but still.I started noticing more when I saw an intel (eg. i7 10700k) open file explorer and other programs in a blink of an eye and feeling overall more snappy that too on a 60fps video, it would feel even more snappier on my 240hz monitor, that's when I decided to switch to intel also the Ram latency on intel is a big reason for me I get around 57ns on 4000MHz @cl18 @1.45v (it was 64ns on stock 3200 c16) Iam currently using windows 11 and I did try windows 10 but it still felt less snappier than I saw on intel.Will the 9700k also have the same snappiness as a 9900k ? because it's probably a 9900k but without HT. I've also heard people saying that 9900k is still snappier than most of the newer CPUs.....I just hope my 9700k won't dissapoint me atleast in terms of overall snapiness and latency

@@HeartOfAdel Iam getting the 9700k in the same price and yes Iam a latency freak and I do notice slightest of the delays aswell, I do have samsung D die right now running in 1:1 at 4000MHz CL18@1.45V I think it would go even higher on intel do you think I should get it then?

@@bazilhussain3378 I can't be responsible for your decision. But if there's no money loss and good ram at hand (provided you're a latency freak) then why not..? I'd still consider only 9900K though to be a clear improvement over 5600x in every way possible. 9700K - really depends

Жарить? 😂 70 градусов при 70-80% нагрузке это хорошая температура для 8-ми ядерного разгонанного процессора. Напряжение и прочее есть в описании. 5ггц ядра 4.8 кольцо едут на 1.24в при макс нагрузке, что в купе с 70-80 градусами очень адекватно. В стресс тестах температура приемлемая, 75-85 в occt large extreme avx2 при 180-200в. В играх немного горячее из-за кульков 3080 FE, но 150-170в это макс при 100% когда там компиляция шейдеров.

И 1.35 это абсолютно небезопасное напряжение с очень быстрой памятью, макс это 1.25-1.3. С каким каким нибудь медленным xmp 3200-3600 cl14, вполне безопасно.

Thank You for watching! :)

I kinda don't. I've mostly learnt through trial and error reading things here and there sporadically. You might want to check out Buildzoid.

Не думаю что будет какой-то большой прорыв, но игры становятся всё требовательнее. В течение ближайших 3 лет думаю будут сильно заметны эффекты скорости/тюнинга этого процессора с картами в лиге 3080+, особенно с FG которая в купе с Reflex начинает создавать цпу-лимиты. Но мощности хватит ещё как я уверен. На канале скоро будет видео сравнение затюненного i9-9900K с тем что на 4.7/4.3 и 3200cl16. Разница иногда сносит крышу, буквальна равна приростам двух поколений.

ruclips.net/video/-Dmhb1QaNbA/видео.html В некоторых играх я заметил что 9900К + 3200мгц 50нс немного ограничивает 3080, к примеру The Division 2 (80-90% гпу на ультра). Разница в нормальных сценариях тоже будет всё таки.

​@@HeartOfAdel Из моего опыта только в киберпанке ощущается иногда недогруз, но там больше дело в игре, мем еще в том, что у меня фпс больше чем у тестеров с ютуба на таких же настройках в таком же спеке но даже с процом лучше чем у меня, но у меня иногда бывает недогруз видяхи, даже со включенным FG, в других играх железно все нормес в которые играл

@@HeartOfAdel Оперативка 4400 с норм таймингами конечно дает свой результат, я с этим не спорю - база

​@@exstiborsch6895ютуберы это вообще другая тема. Самые просматриваемые типа HW и GN вообще как я понимаю все процы тех поколений тестировали с вкл. power limit. Никакого разгона, тюнинга нет и подавно. Бесполезные числа.

Tweaking slopes helped me get 61/63/7/7 stable on this board, try that out. If it doesn't work, I'd sacrifice 100-200mhz for better timings, as they make more impact without requiring higher SA/IO. Sub 35ns DR is only doable on Apex for 10900K. The latency should only go that low on i9-9900K. Just don't bother, that 10900K is already very fast. My board can still pass tests at 4400cl16 62/64/7/7 1.28 SA/IO and even 4400cl15 60/62/7/7 is desktop usable, but some tertiary timings can't go as tight anymore, I found it to be a waste of silicon. Coffee/Comet Lake shouldn't really need anything above 4200mhz / tightest timings.

With some time and care it's possible to boot into windows at 4533mhz cl16 with still low rtl/iol, and with 4400cl15 5ghz ring latency goes to 32.8ns. Absolutely not daily (cause of the board/dimms), but stupidly fast. www.reddit.com/r/overclocking/comments/168v92y/328ns_latency_4x8gb_4400mhz_cl154533mhz_cl16_on/

I just bought a 12700kf to replace my 10900k since it was on sale for $190.00. It is around 20% faster in CPU bound/memory benchmarks comparing tuned vs tuned (51/49 4133 G1 43ns with same b-die used 12700kf) (52/49 4400mhz 36.5ns). Comet lake still holds up. I think DDR5 would actually be slower on Alder lake considering the DDR5 imc is just plain ass. So pretty much: Comet lake -> Alder lake ~20% Comet lake -> Raptor Lake DDR5 7800mhz+ 40% Some numbers if your curious. Tested on a lot of games. 13900k was running 57/45/50 7800mhz 50ns. Some games hate the E-cores on Raptor lake so I switch between that config and 57/5.1 That 9900k of yours should be pretty identical to my 10900k, excluding the 2 cores Ofc and small cache boost.

@@jameyt1 yep thanks for the numbers! Seems about right. I still believe with tuned DDR5 Alder Lake would get another +5-15%. And RP is probably around 40-50% faster on avg. That's still honestly a rather modest increase in cpu performance after the first i9 considering it's been here since 2018. And it seems there's basically been no improvement (if not degradation) in PC snappiness.

@@HeartOfAdel 8khz mice are also more stable on Raptor Lake, probably due to the higher clock speeds. I think the 13900k/14900k does improve snappiness SLIGHTLY as long as e cores are off so it’s not adding any DPC/Process latency. But yeah… not too noticeable. it’s really hard to achieve a perfect polling rate of 8000hz, with my 13900k I get drops to 7000hz still with USB connected to the xHCI slot and all.

Yes, it makes a tremendous impact. It also makes things insanely snappy in windows, something that those chiplets and x3ds still don't have. Full video incoming comparing lazy 4.7/4.3 3200c16 to 5/4.8 4266c16. Spoiler: the difference frequently gets to 30%. RTX 3080 easily stretches its wings like that. That's extremely bad that no one knows how powerful these actually are, thanks to the oblivious out of the box mainstream. People are in complete denial when I tell them fps it can get in some games (like 200 in Warzone), because there's no way to them it can be as fast as the much newer 5700x3d/5800x3d that people think of as a new 1080ti (lol). But who cares, it's their denial. 7700K is still decent with proper tuning, it should easily get to stock ddr4 12400f fps in games I believe.

That's Intel tuning magic :) In all honesty you've just answered your own question - memory and latency (high ring clock should help as well). Xmp slows things down a lot. You'll probably get the same fps after tuning. That map also has a bit better highs than some others, plus I've heard TAAU can help. That gave it a bit extra. It still gets around 200fps avg everywhere else, with destructions dropping things to 160-170. RTX 3080 with i9-9900K are really damn fast in 2023. Thanks for watching! ✊

​@@HeartOfAdel That is crazy. Sadly my ram is c-die and is not able to really overclock even with insane voltages. Best was 55ns... from 60ns stock (And even that improved fps by 5-10fps). You got any recommendations for good ram? Or u think its more worth just waiting and upgrading to ddr5 and 14th gen or 15th later on?

​@@HeadsEnjoyerI think it depends on how long you're going to stay with RTX 3080. If 11700K is fully tuned it's close to 13600/5800x3d in games which is plenty fast for this card for years to come. In that case, finding cheap bdies will be of huge benefit, Asgard Bragi V3 that I own or something else at 3200-3600 cl14. If you plan on getting a newer card, or you're not into fiddling, then it's better to buy something new. 11700K is fast enough even on xmp, so I'd just wait.

Check out the new video. You can see what the gains are in Finals and other games.

hey, try setting nvidia reflex to "On" instead of "On+boost"; Actually makes like 40-50fps difference in the Finals somehow

Thanks bro ✊

Огромное пожалуйста :) очень приятно слышать Warzone как я предполагаю любит потоки, поэтому с 9600К проблемы. Но память решает все, если хорошо затюнить на 3800-4200мгц со всеми ужатами таймингами + хорошая скорость кольца, в Warzone будет 200фпс и 1% должен быть около 140-150. Edit: 200фпс с 9900K конечно, не 9600К.

Shadow Play

понял-спасибо.@@HeartOfAdel

благодарю. ;)

Also one more suggestion is to add frametime graph to your OSD, its quite important for benches :)

I'm not sure I'm gonna return to older games, but the osd will get more detailed in the next videos.

@@HeartOfAdel nice, it would be great if you could at least fit GTA4 or crysis 2007 in there, but yeah looking forward for more videos :)

Yep, absolute mega. Actually I already have added that to the next video with gpu/cpu bottleneck in recent games. Quite a coincidence that you brought it up :)

Благодарю

Ваще