New nForce4 Ultra mobo from ECS
- Thread starter PrayForDeath
- Start date
odd looking board. Looks like a unfinished board(not as many components,resistors,ect). Def looks cheap but I like the looks of the PCI/DIMM slots. But of course it should be much cheaper than most other boards.
The board looks unfinished because the PCB doesn't have any codes written on it. Check this one out:
http://www.ecs.com.tw/ECSWeb/Products/P...il.aspx?DetailID=493&MenuID=90&LanID=0
It has the same components but looks more crowded.
http://www.ecs.com.tw/ECSWeb/Products/P...il.aspx?DetailID=493&MenuID=90&LanID=0
It has the same components but looks more crowded.
There's been a thread a while ago on how ECS suck. Do you think this motherboard will be better than their previous products?
Ah, the parrots again ... there's always been a ton of badmouthing, with very very little actual substance behind it other than "I've heard" and "there's been a thread a while ago". Mostly, that's people searching for an easy scapegoat if their DIY attempt failed - mostly for blatant lack of experience, beginner's mistakes, or plain mechanical clumsiness. Read up on those old threads, BEFORE spinning the myth one more revolution. Thank you.
ECS isn't the world's biggest manufacturer of OEM boards because their stuff sucks. Think about it - in shiny-box retail, if you lose one customer, you lose one mainboard sale. In the OEM business, you lose hundreds of thousands of sales at once if you're having even the slightest quality problem.
ECS isn't the world's biggest manufacturer of OEM boards because their stuff sucks. Think about it - in shiny-box retail, if you lose one customer, you lose one mainboard sale. In the OEM business, you lose hundreds of thousands of sales at once if you're having even the slightest quality problem.
Megatomic
Lifer
- Nov 9, 2000
- 20,127
- 6
- 81
I have only owned one ECS board, it was a D6VAA-R I ran it for quite a while with 2x PIII 1GHz chips and 2GB of RAM. That board was very nice. It's still running today in a buddy's computer.
I wonder if they'll make an nForce Professional 2200 board, I might be able to afford one made by ECS.
I wonder if they'll make an nForce Professional 2200 board, I might be able to afford one made by ECS.
Gamingphreek
Lifer
- Mar 31, 2003
- 11,679
- 0
- 81
ECS never really sucked. They just weren't for enthusiasts as they generally weren't very customizable, and didn't have many frills on them. Additionally there components were kind of cheap. Evidently they are trying to change this impression.
Looks like a nice board though.
-Kevin
Looks like a nice board though.
-Kevin
"kind of cheap" components? Ask "top level" brands like Abit, Epox and ASUS about the capacitor disaster. Then look how many ECS boards with the problem you'll find ...
Originally posted by: Megatomic
Great minds think alike.Originally posted by: CraigRT
haha, hey, check back in the K8AN2E-GR thread
hey you know what else, in that article at FS, the ECS has dual BIOS, which means it probably IS a Gigabyte!
Not many other companies do that is all.. weird!
I didn't mean to say "ECS boards suck", I was just asking if they do. I didn't even own one myself so I can't judge them. It's good to hear they have good mobos for cheap prices. Are they on the same level of Foxconn?
VirtualLarry
No Lifer
- Aug 25, 2001
- 56,571
- 10,206
- 126
Well, it does look pretty spartan on the caps, especially around the DRAM array, although secondarily around the CPU VRMs too. At least they are using three-phase. That might allow them to get away with a smaller size/number of caps. Either way though, definately not a high-end board. I have a hard time believing that it would be stable with all four DIMM slots filled. (My Abit BX6-r2 Slot-1 board had more caps on it, both around the CPU and DRAM array power circuits. Then again, it was designed with overclocking in mind, so it probably had a bit more "headroom", whereas the ECS board is probably designed only for stock speeds, and components selected as such, due to the price-sensitive nature of their market. But the power demands of newer components are higher than back in the Slot-1 days.) Feature-wise, it doesn't look bad at all, and I did spy a few SMT fuses near the USB port headers - that's definately a good sign. Also a decent number of caps between the PCI/PCI-E slots. So it's definately not a bottom-of-the-barrel board. Just a spartan, entry-level (non-enthusiast) one.
Where do you people see "dual-bios" anywhere? I don't see a secondary BIOS chip - it looks like the Phoenix BIOS sticker is actually attached to some other SMT chip, perhaps the board's hardware-monitoring chip or something. The BIOS chip itself isn't even socketed. Definate minus for that, definately a "budget" board. Heaven help you if you screw up the flash.
Edit: Oh, this. The middle picture captioned "dual BIOS", doesn't look like it to me - there's only one surface-mount flash chip in that picture. The reviewer must be clueless. Have FS's reviews gone downhill that badly recently? (larger picture, note that the bottom chip is the flash chip, the top chip is something else, but someone stuck the BIOS's serial-number sticker onto it.)
Wow, three standard IDE ports, and six SATA ports - this thing would make a heck of a base for a low-level file-server or media-server, a duty that the board's non-overclockability wouldn't really affect. (Edit: They do have "overclocking" BIOS settings, my error.)
Here's a pic of the NVidia reference board. Notice the greater number and larger size of the caps used, especially around the CPU VRMs and the slots. If anyone is taking notes, this is one good rule-of-thumb for determining the difference between a "deluxe"/"quality" and "cost-reduced"/"budget" board. Still a little disturbed about the apparent lack of caps near the DRAM array though, even on the reference board, although the ones that are used, are of the higher-quality type. (Is there something about DRAM power-distribution on AMD64 boards that I am unaware of?)
Where do you people see "dual-bios" anywhere? I don't see a secondary BIOS chip - it looks like the Phoenix BIOS sticker is actually attached to some other SMT chip, perhaps the board's hardware-monitoring chip or something. The BIOS chip itself isn't even socketed.
Definate minus for that, definately a "budget" board. Heaven help you if you screw up the flash.Edit: Oh, this. The middle picture captioned "dual BIOS", doesn't look like it to me - there's only one surface-mount flash chip in that picture. The reviewer must be clueless. Have FS's reviews gone downhill that badly recently? (larger picture, note that the bottom chip is the flash chip, the top chip is something else, but someone stuck the BIOS's serial-number sticker onto it.)
Wow, three standard IDE ports, and six SATA ports - this thing would make a heck of a base for a low-level file-server or media-server, a duty that the board's non-overclockability wouldn't really affect. (Edit: They do have "overclocking" BIOS settings, my error.)
Here's a pic of the NVidia reference board. Notice the greater number and larger size of the caps used, especially around the CPU VRMs and the slots. If anyone is taking notes, this is one good rule-of-thumb for determining the difference between a "deluxe"/"quality" and "cost-reduced"/"budget" board. Still a little disturbed about the apparent lack of caps near the DRAM array though, even on the reference board, although the ones that are used, are of the higher-quality type. (Is there something about DRAM power-distribution on AMD64 boards that I am unaware of?)
VIrtualLarry, I believe the BIOS chips get stacked. SO you stack a new BIOS chip on top of the old one if it goes (ECS's so-called "Top-Hat Flash"). I was actually thinking about getting the ECS board as I'm not looking to do any overclocking, but you make a good point about the number of caps. Definitely seems a little on the low side when compared to the reference board. Now you got me thinking...
It's perfectly normal for reference boards to have a higher number of capacitors than retail products.
Looking at other nForce4 Ultra (http://www.watch.impress.co.jp/akiba/hotline/20050129/image/stneo1.html) boards you'll see the number of capacitors around the DIMM area is perfectly normal. If ECS writes the BIOS right I'm sure it won't have a problem having all 4 DIMMs filled.
As far as build quality I wouldn't be concerned at all with this board.
EDIT: since we're talking about capacitors, this board has a metalized one near a yellow PCI slot which is supposed to be used for sound cards, providing higher quality signal. To me that's just marketing though.
The dual-BIOS thing has been explained. It's not the same concept as Gigabyte, it's more like a BIOS recovery tool. A little less convenient but still very useful, and that explains why the BIOS is soldered on the board.
Looking at other nForce4 Ultra (http://www.watch.impress.co.jp/akiba/hotline/20050129/image/stneo1.html) boards you'll see the number of capacitors around the DIMM area is perfectly normal. If ECS writes the BIOS right I'm sure it won't have a problem having all 4 DIMMs filled.
As far as build quality I wouldn't be concerned at all with this board.
EDIT: since we're talking about capacitors, this board has a metalized one near a yellow PCI slot which is supposed to be used for sound cards, providing higher quality signal. To me that's just marketing though.
The dual-BIOS thing has been explained. It's not the same concept as Gigabyte, it's more like a BIOS recovery tool. A little less convenient but still very useful, and that explains why the BIOS is soldered on the board.
That extra capacitor is not for extra _signal_ quality, it's about smooth _power_ to that particular PCI slot. If you've seen the incoming ripple from the average cheapo PSU, and if you've also seen how many cheap-to-midrange TV and sound cards lack proper bulk decoupling themselves, you might actually see an improvement in audio or video noise if you plug that stuff there.
In general, more capacitors doesn't mean better (voltage regulation) quality. Once you've put enough of the right capacitor mix on, extra ones will not make it better anymore, just drive cost up. Reference boards are designed before the actual chips, cost is not an issue, and the properties of the yet-to-be-made chips aren't known at that point. They're - amongst many other purposes - made to figure out how much power and decoupling the final chip actually requires. On CPU regulation in particular, when you're using faster regulators and MOSFETS, you need a lot fewer of the big capacitors. Design choice.
ECS "Top Hat" is a separate BIOS chip on top of an inverted socket. You plug that onto the BIOS chip on the board when you screwed that one's contents up. This is all different to Gigabyte's approach that places two chips onto the board and uses a software (!) switch to use the spare if the primary one is b0rked.
In general, more capacitors doesn't mean better (voltage regulation) quality. Once you've put enough of the right capacitor mix on, extra ones will not make it better anymore, just drive cost up. Reference boards are designed before the actual chips, cost is not an issue, and the properties of the yet-to-be-made chips aren't known at that point. They're - amongst many other purposes - made to figure out how much power and decoupling the final chip actually requires. On CPU regulation in particular, when you're using faster regulators and MOSFETS, you need a lot fewer of the big capacitors. Design choice.
ECS "Top Hat" is a separate BIOS chip on top of an inverted socket. You plug that onto the BIOS chip on the board when you screwed that one's contents up. This is all different to Gigabyte's approach that places two chips onto the board and uses a software (!) switch to use the spare if the primary one is b0rked.
VirtualLarry
No Lifer
- Aug 25, 2001
- 56,571
- 10,206
- 126
Oh, kind of like a "bios-savior"-esque thing? I was kind of wondering what that "Top Hat Flash" logo printed on the board was about. The review was clearly wrong though, suggesting that that 100-odd pin SMT PQFP was somehow a BIOS flash chip. Just because it had the BIOS sticker on that chip doesn't mean anything. I assume that was so that they could put the BIOS revision number sticker on the actual flash chip. So why didn't they show a picture of the "top-hat flash" device? Surely that's not actually included with the board - it would have been far easier and cheaper to simply socket the flash chip in the first place (like the reference board does), rather than engineer and mfg some sort of stacking device with a secondary BIOS chip. (Which one assumes that ECS has to send out in the case of someone's BIOS on their board failing - wouldn't it be far cheaper to simply send out a pre-flashed industry-standard flash chip instead, just to drop into a socket? Then again, ECS is probably saving $0.35/board for not using a socket, which is probably saving them a few $100K in mfg costs, so the additional cost of having to send out 10K of these "top hat flash" devices over the lifetime of the mfg run of the board, is still a financial win overall for them I guess.)Originally posted by: rexxe
VIrtualLarry, I believe the BIOS chips get stacked. SO you stack a new BIOS chip on top of the old one if it goes (ECS's so-called "Top-Hat Flash"). I was actually thinking about getting the ECS board as I'm not looking to do any overclocking, but you make a good point about the number of caps. Definitely seems a little on the low side when compared to the reference board. Now you got me thinking...
Ok, hmm, link link. Ok, I'm wrong - it does appear to be included with the board, but that doesn't make any sense whatsoever, either from a cost perspective, or from a technical usability perspective. What if the original flash chip, actually goes bad. Not just bad programming, but actually fails? Their "top hat flash" solution appears to be useless for that. I know that can happen too - I had a GF2 MX card with a soldered SMT flash chip on it, that had a dodgy BIOS, and I tried to re-flash it several times, and it never really "took", and the one time that I finally was able to get it to do a full-erase, that was the end of the card, I couldn't re-flash it. Up until that point, although I couldn't boot the system with that card, whatever portion of the BIOS contained the internal timing tables for the DRAM and whatnot, that are loaded into the chipset regs by the NV Windows' driver, were still intact, so I could get it to run using the drivers in Windows, as long as I booted off a PCI card. After the erase, it was pretty-much toast. Clearly, something about the flash chip itself (or support circuitry), had gone bad. It does happen.
I admit that ECS's solution is clever, and it looks like they've been able to solder two of those SMT sockets together somehow, while avoiding the cost of including the PCB, but still.. saving the cost of the socket on the mobo, while still including two sockets and two flash chips overall in the package? What in the world? I'd gladly give up their "top hat flash" module, in exchange for a socketed BIOS on the mobo, and a few more caps on the board. I'm guessing that it has to be a marketing thing - so they can put "patented by ECS" on the boxes, as something to differentiate themselves against the other mfgs.
OTOH, I'd kill for one of those devices standalone, I could use them to re-flash some of my other devices with soldered-on SMT flash chips. ECS could have a nifty profitable niche market for those things among computer repair people. Seriously.
VirtualLarry
No Lifer
- Aug 25, 2001
- 56,571
- 10,206
- 126
Absolutely correct.
Someone might want to tell OCWorkbench that though.I guess what they meant to say is "ultra low-noise power quality", which for analog-input devices, voltage drops during high power loads can affect things like the reference voltages used by comparators in things like A/D converters, and thus affect the signals measured by the device.
I mean, it is kind of marketing, to tout that feature out on a single PCI slot - on a quality board, all of the PCI slots should have that level of proper power decoupling, so that the problem should never be an issue, no matter what slot you decide to plug your cards into. But the real world isn't always that kind... I'm a bit miffed at my MSI KT4V-L board for that, precisely because when I did get it, there are nearly zero extra caps between the PCI slots, even though the board layout has all of them clearly marked out, and the pics of the KT4 Ultra model boards that I saw online before my purchase, showed them properly-filled with components. It's kind of egregious because this board has six PCI slots, which implies that it was designed to be filled to the brim with cards. So MSI is just as "bad" in regards to this as ECS - all of the major low-cost/high-volume mobo mfgs tend to have this sort of issue, unfortunately. The number/size of extra decoupling/filtering caps is generally the first thing to go in the process of cost-reducing mass-market boards.
True, but it allows for a greater level of transient power loading before the power to other components starts to degrade, doesn't it? IOW, as long as I don't heavily load the CPU's VRMs, the DRAM array, or the PCI slots, things will be fine, but once I start loading them all down, if the board doesn't have enough power-stability "headroom", everything will start to flake out. Kind of like trying to upgrade a name-brand OEM PC with a low-rated PSU (just enough to cover the pre-installed factory components) with a high-power item, like a 6800 Ultra or something. But in general, the system should run fine using a
I'm guessing that's why the ECS boards offer only limited overclocking features as compared to other "enthusiast" boards, like massive voltage adjustments for vcore, vagp, vdimm, vchipset, etc., because the increased power loading would push the board beyond what it was designed to support, specifically more nominal voltages and components.
I would be surprised if they were designed before actually even recieving test chipset silicon from the mfgs - they have to mfg and do real-world testing/characterization on those test boards before putting them into mass-production, in order to get the bugs out, don't they?
I agree that the reference design is engineered more-or-less independently of the mfg-cost side of the equation. That's sort of the ying and the yang of the mobo business, isn't it? Balancing the designs coming out of engineering with the cost issues of mass-production, and getting a board to market, on time, that balances the two well enough to provide a stable board at a low enough cost point for market acceptance. Given those constraints, I would have to say that ECS actually does pretty well there.
(Forgive me if that seemed a bit presumptuous, I know you work in that biz whereas I don't, but I do know people on engineering staff in other similar industries, and they're always telling me how their engineering designs are being "compromised" by the mfg-cost guys, more or less. I assume that the mobo biz is pretty similar, although cost-reduction is even more critical there, as compared to the defense industry, where 10K units is considered a "massive" production run in many cases.)
That makes some sense. At least they are using three-phase for the CPU, most MSI boards seemingly use two-phase, which is as I understand it a bit harder on the components, and does require bigger caps than a three-phase design. (And then Gigabyte has their over-the-top six-phase/dual three-phase CPU VRM option on some of their boards. That seems almost like flashy overkill to me, although it might not be when dual-core Prescott chips come out, but I doubt they will ever be released for S478.)
I assumed that Gigabyte used a hardware jumper to switch the #CE signals on the chip or something. If they're using software.. that's bizarre. They could have just used a single chip with twice the capacity instead, really. (Maybe a jumper to swap the high address bit instead?)
I wonder how all of these sorts of "dual BIOS" solutions will pan out, when Intel's EFI firmware gets into a major swing? Don't those essentially require you to "upgrade the drivers in your BIOS" for new hardware? Wouldn't that effectively make dual-BIOS solutions more hassle than they are worth? I'd take a socketed BIOS anyday, really.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter