EVGA

Igors Lab: Less MLCC, more crashing and lower clocks?

Page: < 12345.. > >> Showing page 5 of 6
Author
EVGANewMember2719
New Member
  • Total Posts : 71
  • Reward points : 0
  • Joined: 2015/11/18 01:31:24
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:02:47 (permalink)
legit
If it was as simple as 6 MLCCs = Good, why are there quite a number of Asus TUF cards just crashing though?




Source? Who said ASUS TUF cards are crashing?
Yesterdays_news
New Member
  • Total Posts : 100
  • Reward points : 0
  • Joined: 2014/11/04 16:51:30
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:05:18 (permalink)
If ASUS cards are crashing too it must be a driver or BIOS issue which makes sense considering how rushed these cards were.
YURIIII
New Member
  • Total Posts : 48
  • Reward points : 0
  • Joined: 2016/09/20 09:41:07
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:07:32 (permalink)
All I've been hearing through multiple forums is that MSI, Gigabyte and Zotac are crashing. All supposedly with 6+0 or 5+1 configuration.

And just to be clear, I'm not claiming that 100% of them are because of capacitors.
post edited by YURIIII - 2020/09/25 16:10:34
STR8_AN94BALLER
iCX Member
  • Total Posts : 298
  • Reward points : 0
  • Joined: 2015/02/01 22:20:11
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:19:05 (permalink)
Where's Jacob Freeman? This thread needs your answer!
MemphisUFC
New Member
  • Total Posts : 10
  • Reward points : 0
  • Joined: 2016/12/07 17:24:56
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:27:20 (permalink)
h ttps://www.youtube.com/watch?v=_e_KDqzlAbk
 
At the very end of this video i believe he gives a good indication of what may be going on.  Also, very cool feature by ASUS.
justin_43
CLASSIFIED Member
  • Total Posts : 3326
  • Reward points : 0
  • Joined: 2008/01/04 18:54:42
  • Status: offline
  • Ribbons : 7
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:29:30 (permalink)
STR8_AN94BALLER
Where's Jacob Freeman? This thread needs your answer!


You aren't going to get an answer that way. He won't say anything unless there is an official statement from EVGA. I bet they are doing lots of testing right now before they say anything and that's if they are even allowed to by big daddy Nvidia.

ASUS RTX 4090 TUF OC • Intel Core i7 12700K • MSI Z690 Edge WiFi • 32GB G.Skill Trident Z • EVGA 1600T2 PSU
3x 2TB Samsung 980 Pros in RAID 0 • 250GB Samsung 980 Pro • 2x WD 2TB Blacks in RAID 0 • Lian-Li PC-D600WB
EK Quantum Velocity • EK Quantum Vector² • EK Quantum Kinetic TBE 200 D5 • 2x Alphacool 420mm Rads
LG CX 48" • 2x Wasabi Mango UHD430s 43" • HP LP3065 30" • Ducky Shine 7 Blackout • Logitech MX Master
Sennheiser HD660S w/ XLR • Creative SB X-Fi Titanium HD • Drop + THX AAA 789 • DarkVoice 336SE OTL
Yesterdays_news
New Member
  • Total Posts : 100
  • Reward points : 0
  • Joined: 2014/11/04 16:51:30
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:30:51 (permalink)
I wonder how big this will get
Crismac1
Superclocked Member
  • Total Posts : 169
  • Reward points : 0
  • Joined: 2017/04/15 14:13:41
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:32:28 (permalink)
EVGANewMember2719
legit
If it was as simple as 6 MLCCs = Good, why are there quite a number of Asus TUF cards just crashing though?




Source? Who said ASUS TUF cards are crashing?


Since there has been no response, IMHO probably not very legit




 
3080 FTW3 Ultra
Graphics
EVGA 3080 FTW3 Ultra 10G
ROG PG348Q (3440x1440@100Hz)
 EVGA Super Nova 1600 G2 
 1863GB Samsung SSD 970 EVO Plus 2TB 
 1863GB Samsung SSD 970 EVO Plus 2TB
 1863GB Samsung SSD 970 EVO Plus 2TB
 5589GB Western Digital WDC
 1863GB Samsung PSSD T7
https://www.evga.com/member/elite/?referralid=JIMG788TNYTF2DLZB1VX
ThirteenEra
New Member
  • Total Posts : 100
  • Reward points : 0
  • Joined: 2018/10/24 02:30:01
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:35:29 (permalink)
MemphisUFC
h ttps://www.youtube.com/watch?v=_e_KDqzlAbk
 
At the very end of this video i believe he gives a good indication of what may be going on.  Also, very cool feature by ASUS.


I just rewatched second half of stream. I see nothing that qualifies as "indication" of any kind. Can you timestamp please?
 

EDIT: i misunderstood the post, ignore this
degenerate
SSC Member
  • Total Posts : 614
  • Reward points : 0
  • Joined: 2013/08/23 05:38:31
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:43:30 (permalink)
legit
If it was as simple as 6 MLCCs = Good, why are there quite a number of Asus TUF cards just crashing though?




Got a source for that?


 
|5900x|MSI B550 Unify-X|2x16GB G.Skill 3800 C14 @ 1900 FCLK|2TB WD SN850 M.2|EVGA 3090 KPE|EVGA 1600T2|Alienware '25' & LG 48" CX OLED|
guitarwar241
Superclocked Member
  • Total Posts : 104
  • Reward points : 0
  • Joined: 2017/03/15 18:58:09
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 16:47:31 (permalink)
I'd love to have Jacob come and verify some of this information for us. 
dakon
iCX Member
  • Total Posts : 430
  • Reward points : 0
  • Joined: 2006/12/15 20:36:32
  • Status: offline
  • Ribbons : 3
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:00:32 (permalink)
This would also explain why the power slider does not go as high as one would think on evga cards.

12900kf, ROG Strix z690-a Gaming, 32gb Gskill DDR4 4000, EVGA RTX 3090 FTW3 Ultra, Ae-5, Seasonic 1000 Titanium, EK 360 AIO, Corsair 680, LG 38in, AKGK7XX, Samsung NVME 2tb
YURIIII
New Member
  • Total Posts : 48
  • Reward points : 0
  • Joined: 2016/09/20 09:41:07
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:06:08 (permalink)
Nothing that BIOS-update won't fix.

Attached Image(s)

DarkTAO
iCX Member
  • Total Posts : 345
  • Reward points : 0
  • Joined: 2010/12/28 20:14:36
  • Status: offline
  • Ribbons : 1
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:08:37 (permalink)
HawkOculus
MiRai
There's a review up of a 3090 FTW3 Ultra that shows 2x MLCCs and 4x POSCAPs, which differs from the original stock photo shown on sites.
 
Image (Source)
EVGA Store Stock Image
Amazon Stock Image


Great detective work. It seems likely that this is why stock is so low in general. They probably had to scramble at the last minute to change the designs on most of their cards.



Seeing the Asus tough have 6 MLCCs compared to 2 on the 3090 FTW3 Ultra is a little disheartening..
 
Maybe it's a good thing "order placed" failed hundreds of times in the 95 minutes of purgatory. 

 24G-P5-3987-KR
09/24/2020 01:16:10 PM PT Yes
Delivered 17 Hours From Purchase
Intel i9-9900K
EVGA RTX 3090 FTW3 Ultra
EVGA CLC-280
Asus PG279Q
LG CX OLED 55"
EVGA RTX 3090 FTW3 Ultra/Hybrid Cooler
Gigabyte Z390 Aorus Master
32GB G.Skill TridentZ DDR4 3600
EVGA Supernova 850 T2
Samsung 970 EVO 1TB NVMe M.2 SSD
Samsung 850 pro 512GB SSD
Samsung 840 500GB SSD
 
 
08G-P5-3751-KR
10/28/2020 04:27:02 PM PT No


24G-P5-3987-KR
10/02/2020 17:12:43 PM PT No - Not Buying
 Removing Once 3090 Arrives
10G-P5-3897-KR
09/24/2020 01:35:27 PM PT No - Not Buying
Removing Once 3090 Arrives [
kevinc313
CLASSIFIED ULTRA Member
  • Total Posts : 5004
  • Reward points : 0
  • Joined: 2019/02/28 09:27:55
  • Status: offline
  • Ribbons : 22
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:16:03 (permalink)
Frammish
@kevinc313 - you may not like how Lokator said it, and Lokator may or may not know all about capacitors, but he paraphrased the article well and those are the same terms Igor used in the article. Also, there are many different kinds of ceramic and tantalum capacitors. Using the terms he used, Lokator was more specific about the caps than you were. Professionals do call them by their “formal” names when those differences are important to the conversation - such as here with these.
 



You're right, we wouldn't want anyone to get confused here and start putting thru hole parts on their surface mount pcb locations. 
 
For the rest of us, ceramic chip cap (ceramics) and tantalum chip cap (tantulums) works good so people will actually know what you're talking about.
 
And there is a cost difference. I don’t know what prices they are paying but there are 10 MLCCs or one poly in each of the six locations on the back of the board. If there was no cost difference, all boards might be using all MLCC caps like the ROG Strix. Why do you think the cheap boards are using polys? It’s because there is a price difference somehow and the polys are cheaper especially in a 10:1 situation. That’s 10 parts placements instead of one in each location using MLCCs. Maybe not a big difference price wise (you’d need to run all the numbers to know) but the lines will be faster when you use fewer parts. That has a cost and benefit.

 
You should look up the pricing on these parts and the placement speed of a typical SMT placement machine.  I'll wait.  Ceramics are used extensively all over the board and I'd bet that the difference in this case is literally pennies in parts cost and a matter of 2-3 seconds for an array vs. single.  I HIGHLY doubt the decision was driven by cost, but as I noted elsewhere any sane mfg would rather put on one part vs. ten if they can.
 
Regardless, you clearly know what you're talking about and I don't want to get into a navel gazing argument about it.  I'm just looking to refute the narrative that some AIB's are "cheaping out" by selecting some parts vs. others.  Heck, Asus could have had a warehouse full of ceramics to burn through and trouble getting tantulums, or other considerations with their implementation of the reference design where they selected the ceramics.
post edited by kevinc313 - 2020/09/25 18:32:25
kougar
CLASSIFIED Member
  • Total Posts : 3034
  • Reward points : 0
  • Joined: 2006/05/08 10:11:19
  • Status: offline
  • Ribbons : 22
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:25:11 (permalink)
Looks like many companies shipped 3080/3090 cards with 6-poscap designs, then changed them? Glad to see EVGA's changed its XC3 and FTW designs, because every single image on Newegg shows all EVGA 3000-cards with the old 6-poscap design. Same for ASUS, MSI, etc. 
 
According to Igor the official NVIDIA reference spec calls for 1 group of MLCC caps, any AIBs that had designs with none were actually breaking spec. 


Have water, will cool. 
kevinc313
CLASSIFIED ULTRA Member
  • Total Posts : 5004
  • Reward points : 0
  • Joined: 2019/02/28 09:27:55
  • Status: offline
  • Ribbons : 22
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:28:46 (permalink)
kougar
Looks like many companies shipped 3080/3090 cards with 6-poscap designs, then changed them? Glad to see EVGA's changed its XC3 and FTW designs, because every single image on Newegg shows all EVGA 3000-cards with the old 6-poscap design. Same for ASUS, MSI, etc. 
 
According to Igor the official NVIDIA reference spec calls for 1 group of MLCC caps, any AIBs that had designs with none were actually breaking spec. 




Any EVGA close up pic I've seen so far doesn't show signs of surface mount parts being replaced, they came out of the factory with the ceramics in the center.
post edited by kevinc313 - 2020/09/25 18:16:54
INGREDCOLD
Superclocked Member
  • Total Posts : 221
  • Reward points : 0
  • Joined: 2020/09/16 15:09:29
  • Location: Newark Ohio
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:30:46 (permalink)
could there be tuff cards with the earlier cap layout outthere
post edited by INGREDCOLD - 2020/09/25 17:33:48

Ryzen 5800x
X570 Aorus Master
EVGA 3080 FTW3 Ultra
32g Ballistix 3600
2tb Intel 660 500mb WD Black
EVGA 850 Super nova 
EVGA CLC 240mm
PeterFreeman
New Member
  • Total Posts : 48
  • Reward points : 0
  • Joined: 2016/08/26 10:09:31
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:33:34 (permalink)
Its all just speculation right now. You won't get a comment from EVGA yet.
 
Best case scenario its driver related.
jsteedman
New Member
  • Total Posts : 52
  • Reward points : 0
  • Joined: 2018/09/03 20:16:49
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:39:32 (permalink)
kevinc313
 
You should look up the pricing on these parts and the placement speed of a typical SMT placement machine.  I'll wait.  Ceramics are used extensively all over the board and I'd bet that the difference in this case is literally pennies in parts cost and a matter of 2-3 seconds for an array vs. single.  I HIGHLY doubt the decision was driven by cost, but as I noted elsewhere any sane mfg would rather put on one part vs. ten if they can.




jayztwocents put out an update linked on the other page that agrees with this too.  An EE he was talking to went so far as to say MLCCs are cheaper, than SP/POScaps and that you can't necessarily see what type of cap is being used there.  Also that in something like the FTW3s power delivery could be cleaner from more power phases, different voltage regulators, and filtering before it even gets to the chip.
 
https://twitter.com/JayzT...us/1309617232201175040
kevinc313
CLASSIFIED ULTRA Member
  • Total Posts : 5004
  • Reward points : 0
  • Joined: 2019/02/28 09:27:55
  • Status: offline
  • Ribbons : 22
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:41:30 (permalink)
Lokator
MLCC and POSCAP might be similar pricy to buy, but POSCAP might be easier to solder on the board ---> less work time ---> cheaper (yes, I am a noob)

- Asus heavily advertises its fully automated board production process, so for them it might not be an issue to solder more of these smaller MLCC on these boards?

- fully speculated from a noob

oh and nice, Jayz made a video about it, will blow up now I guess



They are all put on by machine on all cards.  Time difference is seconds.  Cost difference is pennies.
kring
iCX Member
  • Total Posts : 274
  • Reward points : 0
  • Joined: 2007/01/19 09:06:42
  • Status: offline
  • Ribbons : 1
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:41:50 (permalink)
my 3090 FTW3 Ultra has 2 in the center like the other pictures posted...

Attached Image(s)

ppcli
New Member
  • Total Posts : 18
  • Reward points : 0
  • Joined: 2019/09/28 21:24:21
  • Location: MB
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:43:44 (permalink)
Thank you, everyone that used bots yesterday when I went to buy a 3090 Ftw3 Ultra and was gone before I could pay for it we’ll it was in the cart.
Karma
post edited by the_Scarlet_one - 2020/09/25 17:59:22


 
INGREDCOLD
Superclocked Member
  • Total Posts : 221
  • Reward points : 0
  • Joined: 2020/09/16 15:09:29
  • Location: Newark Ohio
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:47:57 (permalink)
haven't found anything yet about asus cards crashing. any one got any sources?

Ryzen 5800x
X570 Aorus Master
EVGA 3080 FTW3 Ultra
32g Ballistix 3600
2tb Intel 660 500mb WD Black
EVGA 850 Super nova 
EVGA CLC 240mm
kevinc313
CLASSIFIED ULTRA Member
  • Total Posts : 5004
  • Reward points : 0
  • Joined: 2019/02/28 09:27:55
  • Status: offline
  • Ribbons : 22
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:51:17 (permalink)
Gr3yGhost
Lokator
Will EVGA change it if true?


I hope all the AIB's do so when it comes to creating new batches. It would delay creating new supply to fill the demand. 

I have no idea as to the complexity of the tooling needed to precisely install such small devices, but from what i've seen in some assembly videos - they use automation for the majority of production, they could conceivably re-program the automated machine(s) that specifically deals with installing these to add 1 or more additional groups of these capacitors. 

This does however, increase cost of materials, slows production for a period of time - and causes additional cost to replace anyone who would wish (likely anyone who has this concern) with an updated board. 

The trickle-down effect of making such a production change could be costly. And as someone who runs a business, you want to use the most cost-efficient method as much as possible while also trying to deliver the same quality product. 
Watching Jayz video, I agree with him that the "free" option is a bios update to lower the clocks. 

But I am crossing my fingers, that all AIB's make a change for future production runs of these cards especially on higher tier cards. 

I'll also cross my toes hoping that the 3080 Hybrid has these 2 groups of MLCCs at the very least




Baring any conjecture that this is actually an issue period, to make a running change at the manufacturer is very little effort.  Just a matter of updating the program and loading the parts.  The parts are extremely common. Time difference is seconds.  Cost difference is pennies.
 
The bigger issue is if already built boards have to have the parts replaced manually, per location, going from one tantalum to 10 ceramics, that could take about 10 minutes by a fast expert worker in a plant in Taiwan, or 15-30 minutes by a worker in an RMA center in the US. 
post edited by kevinc313 - 2020/09/25 17:54:42
kevinc313
CLASSIFIED ULTRA Member
  • Total Posts : 5004
  • Reward points : 0
  • Joined: 2019/02/28 09:27:55
  • Status: offline
  • Ribbons : 22
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:57:37 (permalink)
jsteedman
kevinc313
 
You should look up the pricing on these parts and the placement speed of a typical SMT placement machine.  I'll wait.  Ceramics are used extensively all over the board and I'd bet that the difference in this case is literally pennies in parts cost and a matter of 2-3 seconds for an array vs. single.  I HIGHLY doubt the decision was driven by cost, but as I noted elsewhere any sane mfg would rather put on one part vs. ten if they can.




jayztwocents put out an update linked on the other page that agrees with this too.  An EE he was talking to went so far as to say MLCCs are cheaper, than SP/POScaps and that you can't necessarily see what type of cap is being used there.  Also that in something like the FTW3s power delivery could be cleaner from more power phases, different voltage regulators, and filtering before it even gets to the chip.
 
https://twitter.com/JayzT...us/1309617232201175040




Yup.  These filtering caps are just the last stage right before the power goes into the chip.  I mean, they are clearly important, but the overall robustness of the design and implementation plays a big role in how much they are needed.
Mr.Gisoid
New Member
  • Total Posts : 41
  • Reward points : 0
  • Joined: 2017/06/10 22:56:46
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 17:59:34 (permalink)
Lokator
If FTW3 really has 6x POSCAP and Asus Strix 6x MLCC, I really consider waiting for the Strix.

Any Info if FTW3 really has 0x MLCC, 2x less than FE and 6x less than Strix?

Will EVGA change it if true?

I'm waiting for reviews of the FTW3 cards but the Asus Strix 3090 OC reviewed relly well
https://www.techpowerup.com/review/asus-geforce-rtx-3090-strix-oc/
powermix24
Superclocked Member
  • Total Posts : 119
  • Reward points : 0
  • Joined: 2016/12/09 12:45:00
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 18:04:28 (permalink)
EVGANewMember2719
legit
If it was as simple as 6 MLCCs = Good, why are there quite a number of Asus TUF cards just crashing though?




Source? Who said ASUS TUF cards are crashing?


 
Some Asus tuf cards have all poscaps 

Intel 9900K 5GHZ|Windows 10 AE|Gigabyte Z390 Pro Wifi|Corsair H150i ELITE CAPELLIX|16GB GSkill RGB DDR4 3200Mz|EVGA RTX 3080 FTW Ultra|EVGA G3 1000 Watt|EVGA DG-87 Case|Alienware 34'' Curved Gaming Monitor: AW3418DW
kevinc313
CLASSIFIED ULTRA Member
  • Total Posts : 5004
  • Reward points : 0
  • Joined: 2019/02/28 09:27:55
  • Status: offline
  • Ribbons : 22
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 18:12:34 (permalink)
The GN stream just showed a closeup of the 3090 FTW3 they are working with, it has the two ceramic arrays in the middle and they are original.
silentoption
New Member
  • Total Posts : 73
  • Reward points : 0
  • Joined: 2008/06/21 08:51:40
  • Status: offline
  • Ribbons : 0
Re: Igors Lab: Less MLCC, more crashing and lower clocks? 2020/09/25 18:18:08 (permalink)
Did EVGA Update their 3080 FTW 3 with the 2 MLCCs and 4 POSCaps? I saw the Reddit post? 

Case: Thermaltake Core X71 Full Case
Motherboard: ASUS ROG Strix X299-E Gaming II LGA 2066
Processor: Intel i9 10900x Ten Core Processor @ 4.70Ghz
Cooler: Corsair H150i Elite Cappellix 360mm All in One Cooling
Ram: G.SKILL TridentZ RGB Series 64GB (8GBx8) 288-Pin DDR4 SDRAM DDR4 3200
Graphics: EVGA RTX 2080 Super XC
Primary SSD (Primary Boot drive): SAMSUNG 970 Evo Plus 2 TB NVME
Page: < 12345.. > >> Showing page 5 of 6
Jump to:
  • Back to Mobile